You are here

  • Home
  • Archive
  • Volume 71, Issue 6
  • Ethnic differences in anthropometric measures and abdominal fat distribution: a cross-sectional pooled study in Inuit, Africans and Europeans

Article Text

Article menu

Download PDFPDF

Ethnicity & health
Ethnic differences in anthropometric measures and abdominal fat distribution: a cross-sectional pooled study in Inuit, Africans and Europeans
  1. Pernille F Rønn1,2,
  2. Gregers S Andersen1,
  3. Torsten Lauritzen3,
  4. Dirk L Christensen1,4,
  5. Mette Aadahl5,6,
  6. Bendix Carstensen1,
  7. Marit E Jørgensen1,7
  1. 1Steno Diabetes Center, Gentofte, Denmark
  2. 2Department of Public Health, Centre for Arctic Health, Aarhus University, Aarhus, Denmark
  3. 3Department of Public Health, General Practice, Aarhus University, Aarhus, Denmark
  4. 4Department of Public Health, Global Health Section, University of Copenhagen, Copenhagen, Denmark
  5. 5Research Centre for Prevention and Health, the Capital Region of Denmark, Glostrup, Denmark
  6. 6Department of Public Health, Section of Health Services Research, University of Copenhagen, Copenhagen, Denmark
  7. 7National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark
  1. Correspondence to Pernille F Rønn, Steno Diabetes Center, Niels Steensens Vej 2-4, NSK 2.12, Gentofte DK-2820, Denmark; pernille.falberg.roenn{at}regionh.dk

Abstract

Background Ethnic variation in abdominal fat distribution may explain differences in cardiometabolic risk between populations. However, the ability of anthropometric measures to quantify abdominal fat is not clearly understood across ethnic groups. The aim of this study was to investigate the associations between anthropometric measures and visceral (VAT) and subcutaneous abdominal adipose tissue (SAT) in Inuit, Africans and Europeans.

Methods We combined cross-sectional data from 3 studies conducted in Greenland, Kenya and Denmark using similar methodology. A total of 5275 individuals (3083 Inuit, 1397 Africans and 795 Europeans) aged 17–95 years with measures of anthropometry and ultrasonography of abdominal fat were included in the study. Multiple regression models with fractional polynomials were used to analyse VAT and SAT as functions of body mass index (BMI), waist circumference, waist-to-hip ratio, waist-to-height ratio and body fat percentage.

Results The associations between conventional anthropometric measures and abdominal fat distribution varied by ethnicity in almost all models. Europeans had the highest levels of VAT in adjusted analyses and Africans the lowest with ethnic differences most apparent at higher levels of the anthropometric measures. Similar ethnic differences were seen in the associations with SAT for a given anthropometric measure.

Conclusions Conventional anthropometric measures like BMI and waist circumference do not reflect the same amount of VAT and SAT across ethnic groups. Thus, the obesity level at which Inuit and Africans are at increased cardiometabolic risk is likely to differ from that of Europeans.

  • ETHNICITY
  • OBESITY
  • EPIDEMIOLOGY
  • INTERNATIONAL HLTH

https://doi.org/10.1136/jech-2016-207813

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    A growing body of evidence suggests that obesity and the level at which obesity increases the risk of cardiometabolic disease varies with ethnicity.1 ,2 Simple anthropometric measures such as body mass index (BMI) and waist circumference are often used to define overweight and obesity; however, ethnic differences in body size and composition have been identified as a limitation to these measurements.3 The relative distribution of visceral (VAT) and subcutaneous abdominal adipose tissue (SAT) is thought to play a role for the observed difference in cardiometabolic disease between populations.4 Excess VAT has been proposed to be a marker of a dysfunctional SAT layer not being able to expand during caloric excess. The inability of SAT to expand and act as an energy sink will produce a lipid spillover and cause accumulation of lipids in undesired places.5 Furthermore, VAT is suggested to be a better predictor of health risk than general obesity and total body fat.6 However, the relationship between conventional obesity measures and abdominal fat accumulation is not clearly established across ethnic groups. In particular, there is a lack of evidence from indigenous populations in Africa and the Arctic which are undergoing transitions to more westernised lifestyles. This knowledge is important to better identify people at increased health risk and target prevention according to ethnic background.7

    It is known that Arctic populations have high levels of obesity as measured by BMI and waist circumference, but lower levels of lipids, blood pressure, triglyceride, 2-hour glucose and insulin for a given BMI and waist circumference level compared with Caucasians.8 ,9 In comparison, African-Americans seem to have lower amounts of visceral fat for the same BMI compared with Europeans,10–13 but only few comparisons on abdominal fat have been carried out in populations living in the African region.11 ,13 VAT and SAT can be measured by ultrasonography, which has been validated against the gold standard methods MRI and CT,14 but is a more accessible and low cost method for large-scale epidemiological studies.

    The aim of the study was therefore to examine how different anthropometric measures are associated with VAT and SAT in an Inuit, African and European population.

    Methods

    Study population

    The current analyses include cross-sectional data from three studies; the Inuit Health in Transition (IHIT) Study,15 the Kenya Diabetes Study13 and Health2008.16

    The IHIT Study was conducted from 2005 to 2010 as a geographically representative, population-based study among adult Inuit (18+ years) in Greenland which has been described elsewhere.15 The study sample comprised 3253 randomly selected participants and out of those who had a clinical examination, only participants with Inuit ethnicity (n=3083) were included in this study. Ethnicity was determined at enrolment based on the primary language of the participant and self-identification with only one ethnicity allowed for each participant.

    The Kenya Diabetes Study was conducted from 2005 to 2006 among three Kenyan rural populations—the Luo, Kamba and Maasai—and an ethnically mixed urban population living in Nairobi. A detailed description can be found in ref. 13. The definition of the ethnic group was based on parents' ethnicity. At least one parent had to belong to one of the ethnic groups included. The combined rural–urban study population was based on an opportunistic sample with 1459 individuals (1178 rural, 252 urban) aged 17 years and above. A total of 1397 completed the entire clinical examination.

    The Health2008 study was a population-based, cross-sectional study that took place from 2008 to 2009 in Denmark. The study included 2218 randomly selected persons aged 30–60 years residing in the western part of Copenhagen as previously described and 795 chose to participate.16 The invited persons were extracted from the Danish Civil Registration System among persons with Danish citizenship, born in Denmark.

    The studies used the same methodology in data collection and were performed in accordance with the Helsinki Declaration. Ethical approval was obtained from the Ethical Review Committee for Greenland for IHIT, the National Ethical Review Committee in Kenya and the Danish National Committee on Biomedical Research Ethics in Denmark for the Kenya Diabetes Study and from the Ethics Committee of the Copenhagen Region (KA-20060011) for Health2008. Participants gave written or oral informed consent.

    Anthropometric and body composition measurements

    Height and weight were measured with the participants in light clothing and without shoes. Waist circumference was measured midway between the iliac crest and the rib cage on the standing participant and hip circumference at its maximum. BMI was calculated as weight divided by height squared (kg/m2), waist-to-hip ratio as waist circumference divided by hip circumference, and waist-to-height ratio as waist circumference divided by height. Bioimpedance analysis and calculation of body fat percentage were performed on a leg-to-leg Tanita TBF-300MA (Tanita Corporation of Tokyo, Japan or America) in IHIT and Health2008,15 ,17 but was not performed in the Kenya Diabetes Study.

    VAT and SAT were assessed by ultrasonography according to a validated protocol.13 ,14 ,17 ,18 The measurements were performed with a portable ultrasound scanner (Pie Medical) using a 3.5 MHz transducer with the participant lying on their back and at the end of a normal expiration. Both measurements were obtained where the waistline and the abdominal midline meet. The waist was determined as the midpoint between the iliac crest and the lower rib. The distances were measured using electronic calipers placed at three angles: medial, 10 cm left and 10 cm right lateral. VAT was defined as the depth in cm from the peritoneum to the front of the lumbar spine, and derived from a mean of the measures from the three angles. SAT was defined as the depth in cm from the skin to the linea alba determined from one measurement with the transducer placed medical.

    Covariates

    Information on sex and age was retrieved from the civil registration system for the IHIT Study and Health2008, and from personal ID cards or own account in the Kenya Diabetes Study. Smoking status was retrieved from questionnaires and coded as current smoker or non-smoker. Classification of ethnicity as Inuit, African and European in the present study was carried out according to guidelines by Bhopal19 with the concept ethnicity defined as ‘the group a person belongs to as a result of a mix of cultural factors including language, diet, religion and ancestry’.

    Physical activity was measured using a combined accelerometer and heart rate monitor (Actiheart, CamNTech, Cambridge, UK) following the same protocol as described elsewhere.20–22 Participants were asked to wear the monitor for 2–5 days in the Kenya Diabetes Study and the IHIT Study, and for 7 days in Health2008. Physical activity was expressed as physical activity energy expenditure (PAEE) calculated as kJ/kg/day and only participants with >24 hours valid activity data were included. The calibration was either based on individual step tests or group calibration.

    Statistical analyses

    For all variables, multivariate imputations by chained equations method was performed in R software with missing-at-random assumptions (S van Buuren, K Groothuis-Oudshoorn. mice: Multivariate Imputation by Chained Equations in R. J Stat Softw 2011;45). Data were imputed from a ‘rich’ data set containing as many variables as possible to increase the probability that missing values only depend on observed data; hence, to make the missing-at-random assumptions more plausible and reducing bias.23 The obesity variables had around 1–3% missing values, while PAEE had 36% missing. We imputed 50 data sets to obtain sufficiently precise estimates of imputation variability and hence valid inference.23 A mean estimate of the relevant parameters was averaged across the copies and p values were computed according to Rubin's rules.24

    Regression models were used to describe VAT and SAT as functions of anthropometric and body composition measures (BMI, waist circumference, waist-to-hip ratio, waist-to-height ratio and body fat percentage). We used fractional polynomials of the anthropometric measures with powers 1, 2 and ½ to allow for non-linearity in associations.25 The models were tested for sex and ethnicity interactions with the different anthropometric measures. Owing to significant sex interactions and biological differences, all models were stratified by sex. Models with a significant ethnicity interaction were investigated further by plotting the fitted curves separately for each ethnic group. To adjust for potential confounding factors, age, smoking and PAEE were included in the models. Age was included as squared in all models because changes in body composition, especially sarcopenia, begin and advance rapidly after age 50, and hence to improve model fit. The dependent variable was log-transformed in models with skewedness in the distribution of the model residuals. We plotted predicted values of VAT and SAT as functions of anthropometric measures for each ethnic group and separately for sex given median values of the covariates (43 years, smoker and a PAEE of 54 kJ/kg/day).

    In sensitivity analyses, the predictions were repeated for non-smokers, and for a lower PAEE level (40 kJ/kg/day) and a higher (68 kJ/kg/day) to check how variation in these factors across ethnic groups influenced the results. Furthermore, a sensitivity analysis was performed, where participants who did not show up fasting for the examinations or had known diabetes were excluded from the analyses (n=222) to avoid that the ultrasound measures in these people were affected.26 Data management was performed in SAS V.9.4 (SAS Institute, Cary, North Carolina, USA) and statistical analyses in R V.3.1.2 (The R Foundation for Statistical Computing, www.R-project.org).

    Results

    Population characteristics

    Of the 5275 participants included in analyses, 3083 (58.5%) were Inuit, 1397 (26.5%) were African and 795 (15.1%) were European. Table 1 shows the population characteristics. African men and women were significantly younger and had significantly lower median values of BMI, waist circumference, hip circumference, VAT and SAT, compared with the Inuit and Europeans. The prevalence of general obesity (BMI≥30 kg/m2) was 23% in the Inuit, 6.6% in the Africans and 16.1% in the Europeans. The prevalence of central obesity in men (waist circumference >102 cm) was 10.4% in the Inuit, 2% in the Africans and 10.7% in the Europeans, and among women (waist circumference >88 cm) 30.3%, 10.3% and 17.4%, respectively.

    View this table:
    Table 1

    Characteristics of study population by sex and ethnicity (n=5275)

    Visceral adipose tissue

    The predicted values of VAT as a function of the different anthropometric measures are shown in figure 1. Significant ethnicity interactions were found in almost all models. Europeans had the highest VAT levels and Africans the lowest with increasing BMI for both genders (figure 1A). Among men, a BMI of 30 kg/m2 corresponded to a VAT of 10.2 cm (CI 95% 9.6 to 10.8) for Europeans, which was significantly higher than that for the Inuit men (9.2 cm, CI 95% 9.0 to 9.4) and African men (8.4 cm, CI 95% 7.7 to 9.1). Among women, VAT did not differ significantly between Inuit (7.4 cm, CI 95% 7.2 to 7.5) and Europeans (7.7 cm, CI 95% 7.3 to 8.1) at BMI levels of 30 kg/m2, whereas African women had a significantly lower VAT (6.3 cm, CI 95% 5.8 to 6.9). Waist circumference was similarly related to VAT as BMI, but at higher waist circumference values the Inuit women approximated the African slope more than did the European women (figure 1B). Europeans also had a significantly higher VAT accumulation with increasing waist-to-height ratio compared with Inuit and Africans, who had similar VAT levels (figure 1C). The same relation was seen for waist-to-hip ratio (results not shown). In the subanalysis of body fat percentage and VAT including Inuit and Europeans, men had steeper curves than women and a significant ethnic interaction was found for men only with higher levels for Europeans (figure 1D).

    Figure 1
    Figure 1

    Associations between anthropometric measures and visceral adipose tissue for each ethnic group separately for men (left) and women (right) with 95% CI. The regression lines are predicted for a person of 43 years, with a PAEE level of 54 kJ/kg/day and smoker. The p value is for the interaction between ethnicity and anthropometric measure testing the hypothesis of parallel curves between the ethnic groups. The x-axis includes values in the quantile range 5–95% of the different anthropometric measures. PAEE, physical activity energy expenditure.

    Subcutaneous adipose tissue

    In analyses with SAT, we found significant ethnicity interactions with all anthropometric measures (figure 2). In general, women had higher levels of SAT and steeper slopes than men for a given BMI, waist circumference or waist-to-height ratio. In the model with BMI (figure 2A), European and Inuit women had similar slopes for a given BMI, whereas African women had significantly lower levels of SAT. For men, all three ethnic groups differed significantly except for overlapping CIs at high levels of BMI. Waist circumference showed the same pattern, although with a more pronounced ethnic difference for women (figure 2B). The associations of SAT with waist-to-height ratio (figure 2C) and waist-to-hip ratio (results not shown) showed similar ethnic differences with even more pronounced differences for women. In the analysis of body fat percentage and SAT, a statistically significant ethnicity interaction was found for both genders with European men and Inuit women having the highest SAT values (figure 2D).

    Figure 2
    Figure 2

    Associations between anthropometric measures and subcutaneous adipose tissue for each ethnic group separately for men (left) and women (right) with 95% CI. The regression lines are predicted for a person of 43 years, with a PAEE level of 54 kJ/kg/day and smoker. The p value is for the interaction between ethnicity and anthropometric measure testing the hypothesis of parallel curves between the ethnic groups. The x-axis includes values in the quantile range 5–95% of the different anthropometric measures. PAEE, physical activity energy expenditure.

    The association between BMI and body fat percentage

    A significant ethnicity interaction was observed in the relation between BMI and body fat percentage for women but not for men (figure 3), where European women had a significantly higher body fat percentage than Inuit for a given BMI.

    Figure 3
    Figure 3

    Association between body mass index and body fat percentage for each ethnic group separately for men (left) and women (right) with 95% CI. The regression lines are predicted for a person of 43 years, with a PAEE level of 54 kJ/kg/day and smoker. The p value is for the interaction between ethnicity and anthropometric measure testing the hypothesis of parallel curves between the ethnic groups. The x-axis includes values in the quantile range 5–95% of BMI. BMI, body mass index; PAEE, physical activity energy expenditure.

    Sensitivity analyses

    Excluding people, who had known diabetes or showed up non-fasting for the examinations, did not change the results (see online supplementary figures S1–3). Smoking status seemed to influence the associations in models with VAT as functions of BMI and waist circumference; thus, for non-smokers, the modification by ethnicity was less pronounced or disappeared. Non-smoking status did, however, not influence the associations with the other anthropometric measures, and neither in models with SAT. Changing the PAEE level to a lower or higher PAEE did not influence any of the associations.

    Discussion

    In this study of Inuit, Africans and Europeans, we demonstrate significant ethnic and sex differences in abdominal fat distribution. Overall, Inuit and Africans had lower levels of VAT compared with Europeans with ethnic differences most apparent at higher levels of anthropometric measure. SAT levels were generally lower for Inuit and Africans for a given anthropometric measure except among Inuit women who had similar or higher SAT for a given BMI or body fat percentage.

    Comparison with other studies

    We found that African men and women had the lowest levels of VAT for a given BMI or waist circumference. These results are consistent with other studies conducted in African-American men.27–32 The results for women are, however, more mixed with some studies reporting similar VAT levels in African-American women compared with white women.28 ,30 ,31 The current evidence on populations with African ethnicity is mainly based on studies in African-Americans, who differ considerably from the African population included in our study with regard to lifestyle, and may partly explain differences in results. The SAT levels in our study were lowest in Africans for a given anthropometric measure in men and women, which are in contrast with studies reporting higher SAT levels in African-Americans compared with a white population after adjustment for other obesity measures.12 ,31 ,32

    No previous studies have examined anthropometric measures in relation to direct measures of VAT and SAT in Arctic populations, but studies in Inuit populations have observed high levels of central obesity measured by waist circumference which has been linked to lower cardiovascular determinants compared with Europeans.8 ,9 The Inuit men generally had lower VAT and SAT levels than Europeans in the adjusted analyses, whereas the women showed a more mixed pattern with either lower or similar levels despite having the highest absolute values on almost all obesity measures compared with their European counterparts. These results may be explained by differences in the relative leg lengths and body builds in terms of slenderness between populations,33 which may also explain the lower body fat percentage found among Inuit women for the same BMI. Inuit have shorter legs relative to their torso and therefore higher BMI values compared with Europeans.34 In addition, Inuit are stockier, and thus have a lower body fat percentage at given BMI values, compared with slender participants who tend to have less bone mass and connective tissue.33

    The tendency to store fat viscerally and less subcutaneously has been identified in Asian populations and suggested to explain the higher rates of diabetes in Asians compared with Europeans for the same levels of BMI.35 Consequently, universal cut-points for BMI and waist circumference do not apply to all populations and has led to the ongoing debate on whether ethnic-specific cut-points for obesity measures would be more rational.2 ,33 Our study showed a significant ethnic variation in fat distribution and body composition when analysed as continuous measures. Thus, our results confirm the suggested limitations related to the use of BMI and waist circumference, as well as the use of universal cut-points. Other studies have suggested waist-to-hip ratio and waist-to-height ratio as good predictors of health risk.36 Waist-to-hip ratio has been used to assess relative fat distribution; however, a large hip circumference can hide abdominal obesity.36 Furthermore, while waist circumference might overestimate or underestimate risk for tall and short persons, waist-to-height corrects for height and it is possible that a single waist-to-height ratio boundary value can be used across different ethnic, age and sex groups. A waist-to-height ratio of 0.5 has been suggested as a cut-point for cardiometabolic risk and to be superior to BMI and waist circumference.37 In this study, the relations of waist-to-height ratio with SAT showed markedly different values between the ethnic groups and for VAT at higher levels of waist-to-height ratio, which does not support the preferable application of waist-to-height ratio across ethnic groups, at least not to identify VAT and SAT. More studies in different ethnic groups are needed to confirm the applicability of waist-to-height ratio as a predictor of abdominal fat accumulation and cardiometabolic health risk.

    Possible explanations for ethnic differences

    The included studies were conducted in three different countries with very heterogeneous populations and it is possible that some of the variation in body fat distribution is explained by differences in sociocultural, environmental, lifestyle or genetic factors, which we were not able to take into account. We did control for important factors such age, smoking and physical activity, but the lack of comparable data on diet is a limitation. Broadly defined, ethnicity is closely linked with cultural factors including traditional dietary patterns, and some nutritional factors are suggested to contribute to variation in VAT and SAT.38 Also, a number of physiological explanations for ethnic variation in body fat distribution have been proposed, especially related to genetic and epigenetic programming of the tendency of the fat compartments to store lipids.5 Fetal and childhood development may influence the development of obesity in later life39 and it is likely that the populations included in this study may have had a different programming of adipose tissue in early life. Variation in adipose tissue could also be an effect of differences in climate given the geographical variation between the studies, for instance through genetic adaptations to a cold climate to store fuel for heat production.40 In addition, Greenlandic Inuit show genetic and physiological adaptions to a diet rich in Ω-3 polyunsaturated fatty acids, which have large effect sizes on height and weight.41 Thus, the observed ethnic differences in our study are possibly caused by interplay between evolutionary adaptions, physiological, cultural and behavioural factors that are nonetheless measurable in the concept ethnicity.

    Strengths and limitations

    A major strength of the study is the large population with precise measures of VAT and SAT assessed by standardised ultrasonography in all three studies as well as the use of objectively measured physical activity by Actiheart. We used multiple imputation of missing data to avoid excluding participants with missing values. Multiple imputations is considered the optimal method for handling missing data compared with complete-case analysis which may lead to biased estimates of the predictive power of the model and the predictors.23

    The use of ultrasound has some limitations. It captures VAT and SAT at one standardised point, but not the volume. Other measures such as a multisliced MRI provides volume but was not a feasible method in the current studies. Ultrasound is well validated against two-dimensional MRI and CT scans in several studies in different populations.42 ,43

    All studies were population-based and the protocols were developed within a long-standing collaboration on development of large-scale methods for epidemiological studies and all with involvement from the Steno Diabetes Center. However, differences in study design and measurement methods may still exist and could have biased the estimates. We do not have data to rule out bias related to reliability of ultrasound measurements, but interobserver and intraobserver data from the IHIT Study shows a variation of 1.9–5.6%,18 which is in line with the variation of 3–5.4% reported in other studies.14 ,43 ,44

    It is likely that interethnic variation in the studies or rural–urban differences may have influenced the results, though it is most likely in the Kenya Diabetes Study. A rural–urban difference in abdominal obesity has previously been shown in the Kenya Diabetes Study.13 Furthermore, it is possible that differences in general health status could have reduced the comparability between the studies; for instance, the participants in Health2008 were a relatively healthy population.16 Finally, generalising the results from the Kenya Diabetes Study to other African populations should be carried out with caution due to the recruitment strategy. Also, it is of note that due to substantial genetic variation in Africa, the results cannot necessarily be applied to other African populations. Thus, more research in different African populations is warranted.

    In conclusion, simple anthropometric measures do not reflect the same amount of VAT and SAT across different ethnic populations; thus, conventional anthropometric measures and universal cut-points may not reflect the same health risk in Inuit and African populations. These findings have implications for clinical practice when identifying and screening people for cardiometabolic diseases and emphasise the importance of targeting preventive efforts after ethnicity.

    What is already known on this subject

    • Anthropometric measures such as body mass index and waist circumference have limitations partly due to ethnic differences in fat distribution.

    • Ethnic variation in visceral (VAT) and subcutaneous adipose tissue (SAT) may contribute to differences in cardiometabolic risk between populations.

    • The knowledge of abdominal fat and the associated health risk in indigenous populations is scarce.

    What this study adds

    • In general, Inuit and Africans have less VAT and SAT for any given anthropometric measure compared with Europeans.

    • The substantial ethnic variation in VAT and SAT implies that the use of conventional anthropometric measures in clinical practice to identify Inuit and Africans at increased health risk is not optimal and requires more research.

    Acknowledgments

    The authors gratefully acknowledge the participants in all three studies and thank Soren Brage and Kate Westgate, MRC Cambridge University, for their work with preanalyses of the physical activity data. The authors also thank Eva Cecilie Bonefeld-Jørgensen and Centre for Arctic Health, Aarhus University, for their support.

    References

    1. ↵
      1. Tillin T,
      2. Sattar N,
      3. Godsland IF, et al
      . Ethnicity-specific obesity cut-points in the development of type 2 diabetes—a prospective study including three ethnic groups in the United Kingdom. Diabet Med 2015;32:226–34. doi:10.1111/dme.12576
      OpenUrl
    2. ↵
      1. Razak F,
      2. Anand SS,
      3. Shannon H, et al
      . Defining obesity cut points in a multiethnic population. Circulation 2007;115:2111–18. doi:10.1161/CIRCULATIONAHA.106.635011
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Rush EC,
      2. Freitas I,
      3. Plank LD
      . Body size, body composition and fat distribution: comparative analysis of European, Maori, Pacific Island and Asian Indian adults. Br J Nutr 2009;102:632–41. doi:10.1017/S0007114508207221
      OpenUrlCrossRefPubMed
    4. ↵
      1. Nazare JA,
      2. Smith JD,
      3. Borel AL, et al
      . Ethnic influences on the relations between abdominal subcutaneous and visceral adiposity, liver fat, and cardiometabolic risk profile: the International Study of Prediction of Intra-Abdominal Adiposity and Its Relationship With Cardiometabolic Risk/Intra-Abdominal Adiposity. Am J Clin Nutr 2012;96:714–26. doi:10.3945/ajcn.112.035758
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Tchernof A,
      2. Despres JP
      . Pathophysiology of human visceral obesity: an update. Physiol Rev 2013;93:359–404. doi:10.1152/physrev.00033.2011
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Fox CS,
      2. Massaro JM,
      3. Hoffmann U, et al
      . Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation 2007;116:39–48. doi:10.1161/CIRCULATIONAHA.106.675355
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Lear SA,
      2. James PT,
      3. Ko GT, et al
      . Appropriateness of waist circumference and waist-to-hip ratio cutoffs for different ethnic groups. Eur J Clin Nutr 2010;64:42–61. doi:10.1038/ejcn.2009.70
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Jorgensen ME,
      2. Glumer C,
      3. Bjerregaard P, et al
      . Obesity and central fat pattern among Greenland Inuit and a general population of Denmark (Inter99): relationship to metabolic risk factors. Int J Obes Relat Metab Disord 2003;27:1507–15. doi:10.1038/sj.ijo.0802434
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Young TK,
      2. Bjerregaard P,
      3. Dewailly E, et al
      . Prevalence of obesity and its metabolic correlates among the circumpolar Inuit in 3 countries. Am J Public Health 2007;97:691–5. doi:10.2105/AJPH.2005.080614
      OpenUrlCrossRefPubMed
    10. ↵
      1. Kalk WJ,
      2. Joffe BI,
      3. Sumner AE
      . The waist circumference of risk in black South African men is lower than in men of European ancestry. Metab Syndr Relat Disord 2011;9:491–5. doi:10.1089/met.2011.0063
      OpenUrlCrossRefPubMed
    11. ↵
      1. Sumner AE,
      2. Micklesfield LK,
      3. Ricks M, et al
      . Waist circumference, BMI, and visceral adipose tissue in white women and women of African descent. Obesity (Silver Spring) 2011;19:671–4. doi:10.1038/oby.2010.201
      OpenUrlCrossRefPubMed
    12. ↵
      1. Katzmarzyk PT,
      2. Bray GA,
      3. Greenway FL, et al
      . Racial differences in abdominal depot-specific adiposity in white and African American adults. Am J Clin Nutr 2010;91:7–15. doi:10.3945/ajcn.2009.28136
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Christensen DL,
      2. Eis J,
      3. Hansen AW, et al
      . Obesity and regional fat distribution in Kenyan populations: impact of ethnicity and urbanization. Ann Hum Biol 2008;35:232–49. doi:10.1080/03014460801949870
      OpenUrlCrossRefPubMed
    14. ↵
      1. Stolk RP,
      2. Wink O,
      3. Zelissen PM, et al
      . Validity and reproducibility of ultrasonography for the measurement of intra-abdominal adipose tissue. Int J Obes Relat Metab Disord 2001;25:1346–51. doi:10.1038/sj.ijo.0801734
      OpenUrlCrossRefPubMed
    15. ↵
      1. Bjerregaard P
      . Inuit Health in Transition—Greenland Survey 2005–2010. Population sample and survey methods 2010. http://www.si-folkesundhed.dk/upload/inuit_health_in_transition_greenland_methods_5_002.pdf (accessed 20 April 2016).
    16. ↵
      1. Byberg S,
      2. Hansen AL,
      3. Christensen DL, et al
      . Sleep duration and sleep quality are associated differently with alterations of glucose homeostasis. Diabet Med 2012;29:e354–60. doi:10.1111/j.1464-5491.2012.03711.x
      OpenUrlCrossRefPubMed
    17. ↵
      1. Fenger RV,
      2. Gonzalez-Quintela A,
      3. Vidal C, et al
      . Exploring the obesity-asthma link: do all types of adiposity increase the risk of asthma? Clin Exp Allergy 2012;42:1237–45. doi:10.1111/j.1365-2222.2012.03972.x
      OpenUrlCrossRefPubMed
    18. ↵
      1. Jorgensen ME,
      2. Borch-Johnsen K,
      3. Stolk R, et al
      . Fat distribution and glucose intolerance among Greenland Inuit. Diabetes Care 2013;36:2988–94. doi:10.2337/dc12-2703
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Bhopal R
      . Glossary of terms relating to ethnicity and race: for reflection and debate. J Epidemiol Community Health 2004;58:441–5. doi:10.1136/jech.2003.013466
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Christensen DL,
      2. Faurholt-Jepsen D,
      3. Boit MK, et al
      . Cardiorespiratory fitness and physical activity in Luo, Kamba, and Maasai of rural Kenya. Am J Hum Biol 2012;24:723–9. doi:10.1002/ajhb.22303
      OpenUrlCrossRefPubMed
    21. ↵
      1. Dahl-Petersen IK,
      2. Hansen AW,
      3. Bjerregaard P, et al
      . Validity of the international physical activity questionnaire in the arctic. Med Sci Sports Exerc 2013;45:728–36. doi:10.1249/MSS.0b013e31827a6b40
      OpenUrl
    22. ↵
      1. Brage S,
      2. Brage N,
      3. Franks PW, et al
      . Reliability and validity of the combined heart rate and movement sensor Actiheart. Eur J Clin Nutr 2005;59:561–70. doi:10.1038/sj.ejcn.1602118
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. White IR,
      2. Royston P,
      3. Wood AM
      . Multiple imputation using chained equations: issues and guidance for practice. Stat Med 2011;30:377–99. doi:10.1002/sim.4067
      OpenUrlCrossRefPubMed
    24. ↵
      1. Marshall A,
      2. Altman DG,
      3. Holder RL, et al
      . Combining estimates of interest in prognostic modelling studies after multiple imputation: current practice and guidelines. BMC Med Res Methodol 2009;9:57. doi:10.1186/1471-2288-9-57
      OpenUrlCrossRefPubMed
    25. ↵
      1. Royston P,
      2. Ambler G,
      3. Sauerbrei W
      . The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol 1999;28:964–74. doi:10.1093/ije/28.5.964
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Philipsen A,
      2. Carstensen B,
      3. Sandbaek A, et al
      . Reproducibility of ultrasonography for assessing abdominal fat distribution in a population at high risk of diabetes. Nutr Diabetes 2013;3:e82. doi:10.1038/nutd.2013.23
      OpenUrlCrossRefPubMed
    27. ↵
      1. Carroll JF,
      2. Chiapa AL,
      3. Rodriquez M, et al
      . Visceral fat, waist circumference, and BMI: impact of race/ethnicity. Obesity (Silver Spring) 2008;16:600–7. doi:10.1038/oby.2007.92
      OpenUrlCrossRefPubMed
    28. ↵
      1. Hoffman DJ,
      2. Wang Z,
      3. Gallagher D, et al
      . Comparison of visceral adipose tissue mass in adult African Americans and whites. Obes Res 2005;13:66–74. doi:10.1038/oby.2005.9
      OpenUrlPubMedWeb of Science
    29. ↵
      1. Conway JM,
      2. Yanovski SZ,
      3. Avila NA, et al
      . Visceral adipose tissue differences in black and white women. Am J Clin Nutr 1995;61:765–71.
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Despres JP,
      2. Couillard C,
      3. Gagnon J, et al
      . Race, visceral adipose tissue, plasma lipids, and lipoprotein lipase activity in men and women: the Health, Risk Factors, Exercise Training, and Genetics (HERITAGE) family study. Arterioscler Thromb Vasc Biol 2000;20:1932–8. doi:10.1161/01.ATV.20.8.1932
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Hill JO,
      2. Sidney S,
      3. Lewis CE, et al
      . 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 1999;69:381–7.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Camhi SM,
      2. Bray GA,
      3. Bouchard C, et al
      . The relationship of waist circumference and BMI to visceral, subcutaneous, and total body fat: sex and race differences. Obesity (Silver Spring) 2011;19:402–8. doi:10.1038/oby.2010.248
      OpenUrlCrossRefPubMed
    33. ↵
      1. Deurenberg P,
      2. Deurenberg-Yap M
      . Differences in body-composition assumptions across ethnic groups: practical consequences. Curr Opin Clin Nutr Metab Care 2001;4:377–83. doi:10.1097/00075197-200109000-00007
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      1. Charbonneau-Roberts G,
      2. Saudny-Unterberger H,
      3. Kuhnlein HV, et al
      . Body mass index may overestimate the prevalence of overweight and obesity among the Inuit. Int J Circumpolar Health 2005;64:163–9. doi:10.3402/ijch.v64i2.17969
      OpenUrlPubMed
    35. ↵
      1. Sniderman AD,
      2. Bhopal R,
      3. Prabhakaran D, et al
      . Why might South Asians be so susceptible to central obesity and its atherogenic consequences? The adipose tissue overflow hypothesis. Int J Epidemiol 2007;36:220–5. doi:10.1093/ije/dyl245
      OpenUrlAbstract/FREE Full Text
    36. ↵
      WHO. Waist circumference and waist-hip ratio: report of a WHO Expert Consultation. 8–11 December 2008. Geneva: World Health Organisation, 2011.
    37. ↵
      1. Ashwell M,
      2. Gunn P,
      3. Gibson S
      . Waist-to-height ratio is a better screening tool than waist circumference and BMI for adult cardiometabolic risk factors: systematic review and meta-analysis. Obes Rev 2012;13:275–86. doi:10.1111/j.1467-789X.2011.00952.x
      OpenUrlCrossRefPubMed
    38. ↵
      1. Ruttgers D,
      2. Fischer K,
      3. Koch M, et al
      . Association of food consumption with total volumes of visceral and subcutaneous abdominal adipose tissue in a Northern German population. Br J Nutr 2015;114:1929–40. doi:10.1017/S0007114515003682
      OpenUrl
    39. ↵
      1. Cottrell EC,
      2. Ozanne SE
      . Early life programming of obesity and metabolic disease. Physiol Behav 2008;94:17–28. doi:10.1016/j.physbeh.2007.11.017
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      1. Wells JC
      . Ecogeographical associations between climate and human body composition: analyses based on anthropometry and skinfolds. Am J Phys Anthropol 2012;147:169–86. doi:10.1002/ajpa.21591
      OpenUrlCrossRefPubMed
    41. ↵
      1. Fumagalli M,
      2. Moltke I,
      3. Grarup N, et al
      . Greenlandic Inuit show genetic signatures of diet and climate adaptation. Science 2015;349:1343–7. doi:10.1126/science.aab2319
      OpenUrlAbstract/FREE Full Text
    42. ↵
      1. Schlecht I,
      2. Wiggermann P,
      3. Behrens G, et al
      . Reproducibility and validity of ultrasound for the measurement of visceral and subcutaneous adipose tissues. Metab Clin Exp 2014;63:1512–19. doi:10.1016/j.metabol.2014.07.012
      OpenUrl
    43. ↵
      1. De Lucia Rolfe E,
      2. Norris SA,
      3. Sleigh A, et al
      . Validation of ultrasound estimates of visceral fat in black South African adolescents. Obesity (Silver Spring) 2011;19:1892–7. doi:10.1038/oby.2011.213
      OpenUrl
    44. ↵
      1. De Lucia Rolfe E,
      2. Sleigh A,
      3. Finucane FM, et al
      . Ultrasound measurements of visceral and subcutaneous abdominal thickness to predict abdominal adiposity among older men and women. Obesity (Silver Spring) 2010;18:625–31. doi:10.1038/oby.2009.309
      OpenUrlCrossRefPubMed

    Footnotes

    • Twitter Follow Gregers Andersen @gregersandersen

    • Contributors MEJ, DLC and MA designed the studies and collected data; PFR and BC analysed the data; PFR wrote the paper; MA, TL, MEJ, DLC and GSA contributed to interpretation of results and edited the paper; PFR had primary responsibility for the final content. All authors read and approved the final manuscript.

    • Funding The Inuit Health in Transition (IHIT) Study was supported by Karen Elise Jensen's Foundation, NunaFonden, Medical Research Council of Denmark, Medical Research Council of Greenland, and the Commission for Scientific Research in Greenland. Health2008 was supported by the Timber Merchant Vilhelm Bang's Foundation, the Danish Heart Foundation and the Health Insurance Foundation. The Kenya Diabetes Study was supported by DANIDA, Cluster of International Health (University of Copenhagen), Steno Diabetes Center, Beckett Foundation, Dagmar Marshall Foundation, Dr Thorvald Madsen's Grant, Kong Christian den Tiende's Foundation and Brdr. Hartmann Foundation. PFR was funded by the Centre for Arctic Health (Aarhus University).

    • Competing interests None declared.

    • Ethics approval Ethical Review Committee for Greenland, the National Ethical Review Committee in Kenya, the Danish National Committee on Biomedical Research Ethics in Denmark, Ethics Committee of the Copenhagen Region.

    • Provenance and peer review Not commissioned; externally peer reviewed.