Body Mass Index, Waist Circumference and Waist:hip Ratio As Predictors of Cardiovascular Risk-A Review of The Literature

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

European Journal of Clinical Nutrition (2010) 64, 16–22

& 2010 Macmillan Publishers Limited All rights reserved 0954-3007/10 $32.00
www.nature.com/ejcn

REVIEW
Body mass index, waist circumference and waist:hip
ratio as predictors of cardiovascular risk—a review of
the literature
R Huxley1, S Mendis2, E Zheleznyakov2, S Reddy3 and J Chan4

1
Renal and Metabolic Division, The George Institute for International Health, The University of Sydney, Sydney, Australia;
2
Department of Nutrition for Health and Development, World Health Organization, Geneva, Switzerland; 3President, Public Health
Foundation of India, India and 4Department of Medicine and Therapeutics, The Chinese University of Hong Kong, The Prince of Wales
Hospital, Hong Kong, Hong Kong

Overweight and obesity have become a major public health problem in both developing and developed countries as they are
causally related to a wide spectrum of chronic diseases including type II diabetes, cardiovascular diseases and cancer. However,
uncertainty regarding the most appropriate means by which to define excess body weight remains. Traditionally, body mass
index (BMI) has been the most widely used method by which to determine the prevalence of overweight in, and across,
populations as well as an individual’s level of risk. However, in recent years, measures of central obesity, principally waist
circumference and the waist:hip ratio and to a lesser extent the waist:height ratio, which more accurately describe the
distribution of body fat compared with BMI, have been suggested to be more closely associated with subsequent morbidity and
mortality. There is also uncertainty about how these measures perform across diverse ethnic groups; earlier, most of the
evidence regarding the relationships between excess weight and risk has been derived chiefly from Caucasian populations, and
hence, it remains unclear whether the relationships are consistent in non-Caucasian populations. The purpose of this review,
therefore, is to provide an overview of the current evidence-base focusing predominantly on three main questions: (1) Which, if
any, of the commonly used anthropometric measures to define excess weight is more strongly associated with cardiovascular
risk? (2) Which of the anthropometric measures is a better discriminator of risk? and (3) Are there any notable differences in the
strength and nature of these associations across diverse ethnic groups?
European Journal of Clinical Nutrition (2010) 64, 16–22; doi:10.1038/ejcn.2009.68; published online 5 August 2009

Keywords: BMI; cardiovascular disease; waist

Introduction on 466 000 deaths, estimated that optimal survival is


achieved at a BMI of 22.5–25 kg/m2 with reductions in life
It is widely accepted that being overweight, traditionally expectancy of 3 and 10 years in individuals with moderate
defined as having a body mass index (BMI; obtained by (BMI 30–35 kg/m2) and extreme obesity (BMI 40–50 kg/m2),
dividing the individual’s weight in kilograms by height in respectively, the latter being equivalent to the years lost by
metres squared) 425 kg/m2, is a major risk factor for a wide lifetime smoking (Prospective Studies Collaboration, 2009).
range of chronic diseases and injuries including cardio- Although BMI has traditionally been the chosen method
vascular disease (CVD), type II diabetes, and certain site- by which to measure body size in epidemiological studies,
specific cancers including colorectal and breast cancer alternative measures, such as waist circumference (WC) (Wei
(Connolly et al., 2002; Chouraki et al 2008). A recent report et al., 1997; Welborn and Dhaliwal, 2007), waist:hip ratio
from the Prospective Studies Collaboration, which was based (WHR) (Jansses et al., 2004; Bigaard et al., 2005) and the
waist:height ratio (WHtR) (Ho et al., 2003; Ashwell and
Correspondence: Professor R Huxley, The George Institute for International Hsieh, 2005), which reflect central adiposity, have been
Health, University of Sydney, PO Box M201, Missenden Road, Sydney, NSW suggested to be superior to BMI in predicting CVD risk. In
2050, Australia.
part, this stems from the observation that ectopic body fat
E-mail: [email protected]
Received 27 April 2009; accepted 29 May 2009; published online 5 August (i.e. which is stored in the abdomen) is related to
2009 a range of metabolic abnormalities, including decreased
BMI, WC and WHR as predictors of cardiovascular risk
R Huxley et al
17
glucose tolerance, reduced insulin sensitivity and adverse from a non-diabetic state (i.e. normal glucose tolerance
lipid profiles, that are in turn risk factors for type II diabetes or impaired glucose tolerance) to overt type II diabetes was
and CVD. Central adiposity has been highlighted as a explored. The pooled relative risk estimates (95% confidence
growing problem, particularly among Asian populations interval) for incident diabetes associated with a one standard
where individuals may exhibit a ‘normal’ BMI but have a deviation increment in BMI, WC and WHR were 1.87 (95%
disproportionately large WC. Currently, the WHO recognizes CI: 1.67–2.10), 1.87 (95% CI: 1.58–2.20) and 1.88 (95% CI:
that WC between 94.0–101.9 cm in men and 80.0–87.9 cm in 1.61–2.19), respectively, showing that these indicators have
women, and WHR 40.8 and 0.9 in women and men, similar associations with incident diabetes (Table 1). Modest
respectively, correspond with the BMI overweight range of regional differences were reported for WHR (but not with
25–29.9 kg/m2 (WHO, 2000a,b). But, as these estimates are BMI or WC) such that the effect was stronger in Caucasian
derived from predominantly Caucasian populations, it has compared with Asian populations: Europe (1.9, 95% CI:
raised issues about the applicability of these cut-point 1.7–2.2) and United States (1.7, 95% CI: 1.4–2.2) versus
values in non-Caucasian populations (WHO, 2000b). There Asia (1.4, 95% CI: 1.1–1.7).
is no consensus over which of these measures is the most These data are slightly at odds with findings from the
strongly associated with CVD risk, either within or between Obesity in Asia Collaboration (OAC), an individual partici-
different ethnic groups. Providing answers to these funda- pant data meta-analysis involving 4263 000 individuals
mental questions is a key requirement for the effective (73% Asian) from 21 cross-sectional studies in the Asia-
management of weight and for defining prevention Pacific region (Huxley et al., 2008). Findings from this study
strategies for the weight-related morbidity within and indicated that with the exception of Caucasian men,
between populations. measures of central obesity were actually more strongly
Hence, the purpose of this report was to provide an associated with prevalent diabetes than BMI (Huxley et al.,
overview of the current literature focusing on three main 2008). For example, a 0.5 standard deviation increment in
questions: (1) Which, if any, of the commonly used anthro- BMI was associated with a 20–30% prevalent odds ratio
pometric measures to define excess weight is more strongly of diabetes, whereas for WC and WHR the same
associated with CVD risk? (2) Which of the anthropometric standard increment was associated with about 40% risk of
measures is a better discriminator of CVD risk? (3) Are there diabetes (Figures 1a and b). The same, however, was not true
any notable differences in the strength and nature of these for hypertension; for a standard increment, the odds of
associations across diverse ethnic groups? hypertension were comparable across the three anthropo-
metric measures for both men and women, although of note
was the stronger association in Caucasians compared with
Association between measures of global and central obesity with non-Caucasian populations. For example, a 0.5 s.d. incre-
hypertension, diabetes and dyslipidaemia ment in each of the three measures of current body size
Over the past two decades, several hundred papers have been was associated with a 40% risk of prevalent hypertension
published that have reported on some aspect of the in Caucasian men compared with only a 30% risk in non-
association between different measures of current body size Caucasian men (Figures 1a and b).
and one or other cardiovascular risk factors. Several authors In a comparable meta-analysis from the Diabetes
have attempted to systematically evaluate the strength and Epidemiology: Collaborative Analysis of Diagnostic Criteria
nature of these associations and it is these overviews that in Asia Study (DECODA, 2008), which involved the collation
form the basis of this current review. Vazquez and colleagues of data from 16 cross-sectional studies, an examination of
conducted a meta-analysis of cohort studies that examined the strength of association between BMI, WC, WHR and
the association between different anthropometric measures WHtR with type II diabetes suggested little difference
of obesity and risk of incident type II diabetes (Vazquez et al., between the first of the three measures but a slightly
2007). In all 32 of the included studies, the progression stronger association with WHtR in both men and women:

Table 1 Pooled relative risk for BMI, WC and WHR with incident diabetes stratified by age, gender and geographical region

Measurement No. of studies Pooled relative risk

Overall Age group Gender Region

Total 32 Overall o50 years X50 years F M Asia US Europe

Body mass index 32 1.87 1.7 2.0 2.4 2.0 2.4 1.7 2.0
Waist circumference 18 1.87 1.6 2.0 2.3 2.9 2.4 1.9 2.1
Waist : hip ratio 25 1.88 2.1 1.7 3.0 2.7 1.4 1.7 1.9

Abbreviations: F, female; M, male.


Adapted from Vazquez et al., (2007).

European Journal of Clinical Nutrition


BMI, WC and WHR as predictors of cardiovascular risk
R Huxley et al
18
P-values for
MEN heterogeneity

Hypertension Odds Ratio BMI vs Asian vs


(95% CI) Other Caucasian
Asian
BMI 1.41 (1.37 –1.45) 0.001
Waist 1.39 (1.35 –1.43) 0.48 0.021
Waist:Hip 1.34 (1.28 –1.39) 0.046 0.29
Caucasian
BMI 1.29 (1.24 –1.35)
Waist 1.28 (1.20 –1.37) 0.85
Waist:Hip 1.29 (1.23 –1.36) 0.99

Diabetes
Asian
BMI 1.26 (1.20 –1.33) 0.006
Waist 1.35 (1.28 –1.43) 0.057 0.20
Waist:Hip
1.47 (1.35 –1.60) 0.002 0.44
Caucasian
BMI 1.39 (1.33 –1.46)
Waist 1.42 (1.36 –1.50) 0.49
Waist:Hip 1.41 (1.33 –1.50) 0.63

1 1.2 1.4 1.6 1.8


Odds Ratio (95% CI)

P-values for
heterogeneity
WOMEN
Odds Ratio BMI vs Asian vs
Hypertension
(95% CI) Other Caucasian
Asian
1.33 (1.28 –1.39) <0.001
BMI
1.37 (1.31 –1.43) 0.33 0.02
Waist
1.25 (1.20 –1.30) 0.034 0.29
Waist:Hip
Caucasian
1.22 (1.20 –1.25)
BMI
1.24 (1.20 –1.29) 0.44
Waist
1.20 (1.16 –1.24) 0.41
Waist:Hip

Diabetes
Asian
1.23 (1.19 –1.28) 0.004
BMI
1.40 (1.32 –1.47) <0.001 0.042
Waist
1.40 (1.29 –1.52) 0.006 0.006
Waist:Hip
Caucasian
1.32 (1.28 –1.37)
BMI <0.001
1.50 (1.44 –1.58)
Waist <0.001
1.62 (1.52 –1.72)
Waist:Hip

1 1.2 1.4 1.6 1.8


Odds Ratio (95% CI)

European Journal of Clinical Nutrition


BMI, WC and WHR as predictors of cardiovascular risk
R Huxley et al
19
age-adjusted odds ratios for diabetes in men (women) for one have reported on the relationship between general and
standard deviation increment were 1.52 (1.59), 1.54 (1.70), central obesity with cardiovascular outcomes.
1.53 (1.50) and 1.62 (1.7), respectively. For hypertension, the
findings from DECODA were comparable with those from Association between obesity indices and CVD risk
the OAC such that there was little evidence that measures of A review of the published evidence indicates that there is no clear
central obesity were more strongly associated with hyperten- agreement as to whether measures of central obesity are more
sion: the prevalent odds ratios for hypertension were 1.68 strongly associated with cardiovascular morbidity and mortality
(1.55), 1.66 (1.51), 1.45 (1.28) and 1.63 (1.5). compared with BMI, and there is a clear need for further long-
The relationship between measures of body anthropo- term, large cohort studies to examine this issue further.
metry with dyslipidaemia, and its individual lipid compo- The Asia Pacific Cohort Studies Collaboration (Asia Pacific
nents, has been less widely studied. The OAC has recently Cohort Studies Collaboration 2006) comprises data from
conducted the most comprehensive series of analyses 440 cohort studies within the Asia-Pacific region. Of these
to date of the relationships between total cholesterol, studies, 33 cohorts (n ¼ 310 000 individuals) had information
high-density lipoprotein cholesterol, low-density lipoprotein on BMI and cardiovascular events but only six cohorts
cholesterol and triglycerides with measures of global and (n ¼ 45 998) had information on waist and HC. In this
central obesity in Asian and non-Asian populations (Barzi subgroup analysis, which was based on 601 coronary heart
et al., in press). There were several key findings from this disease events and 346 strokes, a one standard deviation
study; first, the magnitude of the associations between increase in BMI, WC, HC and WHR was associated with an
measures of body size and lipids were broadly similar increase in risk of CHD of 17% (95% CI: 7–27%), 27% (95%
between Asians and non-Asians. Second, no single measure CI: 14–40%), 10% (95% CI: 1–20%) and 36% (95% CI:
of body size was superior at discriminating those individuals 21–52%), respectively. Subgroup analysis indicated that
at increased risk of dyslipidaemia and, finally, WHR cut- these associations were stronger in those aged o65 years,
points of 0.8 in women and 0.9 in men, in both sexes, were in men and in the non-Asian cohorts; however, caution
applicable across both regions for the optimal discrimination should be applied when interpreting these analyses given the
of individuals with any form of dyslipidaemia in line with relatively small number of events within the subgroups and
previous findings from this collaboration that showed that the overlapping confidence intervals around the point
these values are also optimal for the discrimination of estimates. The authors further concluded that the associa-
individuals with diabetes and hypertension (Huxley et al., tions tended to be consistently stronger for WC and WHR
2008). and weakest for HC by comparing the change in the
There are, however, several limitations of the data from likelihood ratio w statistic (which is used as a measure of
both the OAC and DECODA groups. First, these analyses are the improvement in the goodness of fit of the model)
cross-sectional, which precludes examination of the temporal between the indices; but it should be noted that the
nature of the association between measures of excess weight differences in the likelihood ratio were modest (e.g. 276 for
and cardiovascular risk factors, which is potentially of WHR versus 271 for WC) and hence the clinical relevance
concern given that the development of diabetes or hyper- is questionable. By comparison, there was no clear asso-
tension may influence body size. Second, these reviews have ciation between any of the anthropometric indices with
been limited to examining the association between measures stroke outcomes; a one standard deviation increase in BMI,
of body size and surrogate measures of cardiovascular risk WC, HC and WHR was associated with a hazards ratio of
rather than between morbidity and mortality. This is largely 1.03 (95% CI: 9 to 16%), 1.05 (95% CI: 9 to 20%), 0%
as a consequence of there being far fewer data available on (95% CI: 11 to 13%) and 9% (95% CI: 8 to 28%),
the relationship between different measures of adiposity and respectively. Furthermore, this study did not examine what
mortality outcomes, largely because it has not been until happens to the relationship between BMI and CVD risk
relatively recently that investigators have started to record if adjustment is made for central obesity, which would
measures of central obesity in their studies. Again, this is address the issue of whether the effects of BMI on risk are
because of the greater difficulty, both in practical and independent of central obesity.
cultural terms, in measuring waist and hip circumference This question was explored by INTERHEART, a large case–
(HC) as opposed to weight and height. Below, we have control study involving 412 000 cases of myocardial infarc-
summarized the data from some large-scale overviews that tion (MI) and 14 000 controls of varying ethnicity from

Figure 1 Age-adjusted odds ratios and 95% confidence intervals for prevalent type II diabetes and hypertension associated with 0.5 s.d.
increment in each anthropometric measure: body mass index (BMI), waist circumference (WC) and waist:hip ratio (WHR). Results are shown
separately by sex (a, for men; b, for women) and ethnic group (Asian, Caucasian). The strength of the association between WC and diabetes or
hypertension and between WHR and diabetes or hypertension are compared against the strength of the association between BMI and diabetes or
hypertension. For each variable, the strength of the association with diabetes or hypertension is compared between Asian and Caucasian
individuals. P-values for the differences are shown. Figure 1(a, b) is reproduced through kind permission of Wiley–Blackwell (Huxley et al. 2008).

European Journal of Clinical Nutrition


BMI, WC and WHR as predictors of cardiovascular risk
R Huxley et al
20
52 countries (Yusuf et al., 2005). In this study, BMI was Table 2 Association between an increase in WC and WHR and
positively and linearly associated with MI such that cardiovascular disease risk in men and women (after minimal adjustment
for age and cohort characteristics)
individuals in the top quintile of the BMI distribution had
an approximately 40% greater risk of MI compared Increase in CVD Waist circumference (cm) Waist:hip ratio (U)
with those in the lowest quintile: odds ratio 1.44, 95% CI: risk (%)
1.32–1.57. After adjusting for WHR, the risk was significantly M F Combined M F Combined
attenuated such that for the same comparison, the risk of
10 4.71 5.08 5.04 0.02 0.02 0.02
MI was reduced to approximately 10%: odds ratio comparing
20 9.02 9.72 9.65 0.03 0.04 0.03
highest with the lowest quintile of BMI 1.12 (95% CI: 30 12.98 13.99 13.88 0.05 0.05 0.05
1.03–1.22). As with BMI, WC and WHR were also strongly 40 16.64 17.95 17.80 0.06 0.07 0.06
and linearly associated with risk of MI, but unlike BMI, the 50 20.06 21.63 21.64 0.08 0.08 0.08
relationships were relatively unaffected after adjustment was Adapted from de Koning et al., (2007).
made for BMI, indicating the independence of measures of
central obesity in predicting risk of MI. In models adjusted
for age, sex, region and smoking, the odds ratio for MI in both men and women (an alternate way of viewing the
comparing the top with the lowest quintiles for WHR and data is that a 1 cm increase in WC and a 0.01 increase in
WC were 1.75 and 1.33, respectively (both P-values o0.001), WHR was associated with a 2 and 5% increased risk of
indicating a stronger association between WHR and risk of incident CVD, respectively; Table 2). Further adjustment for
subsequent MI compared with WC. smoking and lipids had no material effect on the results,
The observation from INTERHEART that WHR is more indicating the independent nature of the relationship
strongly associated with cardiovascular risk compared with between measures of central obesity and CVD risk. However,
BMI or WC is at odds with findings from a recent study that this review had two notable limitations; first, it was unable
involved a combined analysis of the Physician’s Health Study to compare the strength of the association between measures
(n ¼ 16 221 men) and the Women’s Health Study (n ¼ 32 700) of general and central obesity with CVD because it did not
(Gelber et al., 2008). In this study of 41900 CVD events include studies that had also reported on the association
(22% in women), which compared the cardiovascular risk between BMI and CVD. Second, although the authors stated
associated with self-reported anthropometric indices (BMI, in the review that they compared the strength of association
WC, WHR and WHtR), linear and positive associations were of WC and WHR with CVD risk by pooling risk estimates
shown between each of these indices with CVD risk, the comparing the highest versus the lowest quantiles of WC
magnitude of which was broadly similar across the measures. and WHR, this is not strictly statistically correct as it would
There was some evidence that, especially in men, the WHtR have required that the analysis be restricted to those studies
was more strongly associated with CVD risk (and WHR the that had reported on both WC and WHR, which was not
least strongly associated);however, after adjusting for BMI, the case.
the relationship was attenuated, but remained statistically
significant. For example, the adjusted hazard ratio for CVD
in men with WHtR X0.69 was 2.36 (95% CI: 1.61–3.47) Which anthropometric measure is the better discriminator of
compared with those with 0.49 o WHtR p0.53 and after cardiovascular risk?
adjustment for BMI the HR was reduced to 1.73 (95% CI: Given the broad similarities in the magnitude of the
1.05–2.83). A similar effect was also shown when BMI was relationship between different measures of current body size
added to WC, suggesting that some of the risk associated with cardiovascular risk and its risk factors, it is perhaps not
with central obesity is mediated in part by BMI. The authors surprising that the discriminatory capability of each of these
concluded that although WHtR tended to be more strongly measures, as assessed by the area under the receiver
associated with CVD risk compared with BMI, the actual operating characteristic curve, at identifying those indivi-
difference between the measures was small and unlikely to duals with the highest cardiovascular risk is also comparable.
be clinically meaningful. The OAC reported on the ability of BMI, WC and WHR to
De Koning and colleagues conducted a meta-analysis of discriminate those individuals with prevalent diabetes or
studies that had reported on the association between WC hypertension and showed that the area under the receiver
and/or WHR with cardiovascular outcomes (de Koning et al., operating characteristic curves ranged from 0.63 to 0.71 in
2007). A total of 15 cohort studies with information on men and from 0.66 to 0.80 in women with little statistically
4250 000 individuals and 4355 CVD events were eligible for significant evidence of any consistent difference between the
inclusion. Eight of these cohorts had reported on the three measures across the sex and ethnic groups.
relationship between WHR and WC with CHD, four on Lee et al. (2008) conducted a meta-analysis involving 10
WHR (only) with CVD (either stroke or CHD) and three on studies (nine of which were cross-sectional) and over 88 000
WC (only) and CVD outcomes. In a minimally adjusted individuals, to determine which of the four indices (BMI,
model, a 10% increase in CVD risk equated to an approxi- WC, WHR and WHtR) is the best discriminator of major
mately 5% increase in WC and a 0.02 unit increase in WHR cardiovascular risk factors: hypertension, type II diabetes and

European Journal of Clinical Nutrition


BMI, WC and WHR as predictors of cardiovascular risk
R Huxley et al
21
Table 3 Comparison of the discriminatory power (pooled AUC score) for three cardiovascular risk factors between measurements of obesity (BMI, WC,
WHR, WHtR) stratified by gender (Lee et al., 2008)

CV risk factors Hypertension (n ¼ 8) Type II diabetes (n ¼ 9) Dyslipidaemia (n ¼ 7)

Measurements Men Women Men Women Men Women

Body mass index 0.64 0.69 0.67 0.69 0.65 0.64


Waist circumference 0.67 0.71 0.70 0.74 0.64 0.66
Waist:hip ratio 0.67 0.71 0.72 0.75 0.64 0.66
Waist:height ratio 0.68 0.73 0.73 0.76 0.67 0.68

dyslipidaemia. In both men and women, measures of central prevalent diabetes associated with a 0.5 s.d. increment in
obesity were superior to BMI as discriminators of cardio- each of the three indices of body weight with prevalent
vascular risk factors, although the differences were small diabetes were consistently stronger in Caucasians. By
and unlikely to be of clinical relevance (Table 3). Further, the comparison, for the same standard increment in anthropo-
study showed that combining BMI with any measure of metric indices, the odds of hypertension were stronger
central obesity did not improve the discriminatory capability (although not always statistically significantly so) in Asians
of the individual measures. compared with Caucasians for both men and women
(Figures 1a and b). Findings from the APCSC substudy of
six longitudinal cohorts showed that the strength of the
Ethnic differences in association between anthropometric associations between BMI, WC, WHR and HC with cardio-
measures and CVD risk vascular risk was similar in the Asian and non-Asian cohorts.
Recently, evidence has accumulated to suggest that the However, as discussed earlier, these analyses are based on a
increasing prevalence of type II diabetes and CVD in Asian relatively small number of events and require validation
countries is occurring at levels of BMI much lower than the from future prospective studies.
WHO BMI cut-point of 25.0 kg/m2. One potential explana-
tion that has been suggested to explain the diabetes
epidemic across large parts of Asia is that ethnic differences Summary
may exist in the strength of the relationships between body
size and metabolic and cardiovascular risk factors. For This review attempted to summarize the evidence for three
example, several studies have shown that, for a given BMI, main questions. The first of these asked if there was evidence
adiposity can be substantially greater in Asian compared to indicate which of the commonly used measures to assess
with Caucasian individuals. Moreover, there is evidence to body size is more strongly associated with subsequent
suggest that within Asian populations there is significant cardiovascular risk. In totality, the evidence was conflicting;
variation in the association between adiposity and BMI. For for diabetes, there was some evidence to indicate that
example, Hong Kong Chinese, Indonesians, Singaporeans measures of central obesity were more strongly associated
and urban Thai have been shown to have lower BMI’s at a with risk compared with BMI, but this was not the case for
given percentage of body fat compared with Europeans, hypertension and dyslipidaemia where the relationships
whereas individuals from Northern China (Beijing) and rural with BMI, WC and WHR were similar. For cardiovascular
Thailand had similar values to Europeans (Deurenberg and outcomes, the evidence again was conflicting, with most
Deurenberg-Yap, 2003). Further studies have reported ethnic studies (with the notable exception of INTERHEART),
differences in the slopes of the associations between BMI and suggesting that the magnitude of the relationships between
CVD risk factors. For example, Bell and colleagues observed BMI and central obesity with cardiovascular mortality is
a stronger association between BMI and hypertension in broadly consistent. However, much of the evidence is based
Chinese compared with Caucasians, and in non-Hispanic on cross-sectional studies and there is a clear need for further
Blacks compared with Caucasians and Mexican Americans data from large-scale longitudinal studies. Perhaps not
(Bell et al., 2002). Similarly, the relationship between body surprisingly, given the general consistency in associations
build with fasting insulin concentration has been shown between measures of body size and cardiovascular risk, there
to be significantly steeper in South Asian compared with was limited evidence to support the superior discriminatory
Caucasian children (Whincup et al., 2002). capability of any of the measures. Furthermore, the differ-
Data from the OAC suggested that there was no evidence ences in discriminatory capability that were reported were
that the strength of the associations between BMI, WC or too small to be of any clinical relevance. Finally, despite the
WHR and diabetes were stronger in Asians compared with often considerable differences in body size and fat distribu-
Caucasians in both sexes (Figures 1a and b). Rather, the tion between different ethnic groups, there was little
reverse was true, particularly in women, where the odds of evidence to indicate that the magnitude of the associations

European Journal of Clinical Nutrition


BMI, WC and WHR as predictors of cardiovascular risk
R Huxley et al
22
between measures of body size and subsequent risk was obesity indicators in relation to diabetes and hypertension in
appreciably different. However, again, the evidence is largely Asians. Obesity (Silver Spring) 16, 1622–1635.
Deurenberg P, Deurenberg-Yap M (2003). Validity of body composi-
cross-sectional and requires confirmation from prospective
tion methods across ethnic population groups. In: Modern aspects
studies. of nutrition: present knowledge and future perspectives. Forum
Nutr Basel Karger 56, 299–301.
Gelber RP, Gaziano JM, Orav EJ, Manson JE, Buring JE, Kurth T
Conflict of interest (2008). Measures of obesity and cardiovascular risk among men
and women. J Am Coll Cardiol 52, 605–615.
Ho SY, Lam TH, Janus ED (2003). The Hong Kong Cardiovascular
The authors declare no conflict of interst. Risk Factor Prevalence Study steering committee. Waist to stature
ratio is more strongly associated with cardiovascular risk factors
than other simple anthropometric indices. Ann Epidemiol 13,
Acknowledgements 683–691.
Huxley R, James WP, Barzi F, Patel JV, Lear SA, Suriyawongpaisal P
et al. (2008). Obesity in Asia Collaboration. Ethnic comparisons of
Rachel Huxley is funded by a Career Development Award the cross-sectional relationships between measures of body size
from the National Heart Foundation of Australia. with diabetes and hypertension. Obes Rev 9 (Suppl 1), 53–61.
Jansses I, Katzmarzyk PT, Ross P (2004). Waist circumference and not
body mass index explains obesity-related health risk. Am J Clin
Nutr 79, 379–384.
References Lee CM, Huxley RR, Wildman RP, Woodward M (2008). Indices of
abdominal obesity are better discriminators of cardiovascular risk
Ashwell M, Hsieh SD (2005). Six reasons why the waist-to-height factors than BMI: a meta-analysis. J Clin Epidemiol 61, 646–653.
ratio is a rapid and effective global indicator for health risks of Prospective Studies Collaboration (2009). Body-mass index and
obesity and how its use could simplify the international public cause-specific mortality in 900 000 adults: collaborative analyses
health message on obesity. Int J Food Sci Nutr 56, 303–307. of 57 prospective studies. Lancet 363, 1083–1096.
Asia Pacific Cohort Studies Collaboration (2006). Central obesity and Vazquez G, Duval S, Jacobs Jr DR, Silventoinen K (2007). Comparison
risk of cardiovascular disease in the Asia Pacific region. Asia Pac J of body mass index, waist circumference, and waist/hip ratio in
Clin Nutr 15, 287–292. predicting incident diabetes: a meta-analysis. Epidemiol Rev 29,
Barzi F, Woodward M, Czernichow S, Lee CMY, Kang JH, Janus E et al. 115–128.
(in press). The discrimination of dyslipidaemia using anthropo- Wei M, Gaskill SP, Haffner SM, Stern MP (1997). Waist circumference
metric measures in ethnically diverse populations of the Asia– as the best predictor of noninsulin dependent diabetes mellitus
Pacific region: the Obesity in Asia Collaboration. Obes Rev. (NIDDM) compared to body mass index, waist/hip ratio and other
Bell C, Adair LS, Popkin B (2002). Ethnic differences in the asso- anthropometric measurements in Mexican Americans–a 7-year
ciation between body mass index and hypertension. Am J prospective study. Obes Res 5, 16–23.
Epidemiol 155, 346–353. Welborn TA, Dhaliwal SS (2007). Preferred clinical measures
Bigaard J, Frederiksen K, Tjønneland A, Thomsen BL, Overvad K, of central obesity for predicting mortality. Eur J Clin Nutr 61,
Heitmann BL et al. (2005). Waist circumference and body 1373–1379.
composition in relation to all-cause mortality in middle-aged Whincup PH, Gilg JA, Papacosta O, Seymour C, Miller GJ, Alberti KG
men and women. Int J Obes (Lond) 29, 778–784. et al. (2002). Early evidence of ethnic differences in cardiovascular
Chouraki V, Wagner A, Ferrières J, Kee F, Bingham A, Haas B et al. risk: cross sectional comparison of British South Asian and white
(2008). Smoking habits, waist circumference and coronary artery children. BMJ 324, 635.
disease risk relationship: the PRIME study. Eur J Cardiovasc Prev World Health Organization (2000a). Obesity: Preventing and Mana-
Rehabil 15, 625–630. ging the Global Epidemic. WHO Technical report series No. 894.
Connolly BS, Barnett C, Vogt K, Li T, Stone J, Boyd N (2002). A meta- WHO: Geneva.
analysis of published literature on waist-to-hip ratio and risk of World Health Organization. International Association for the Study
breast cancer. Nutr Cancer 44, 127–138. of Obesity, International Obesity Task Force (2000b). Health
de Koning L, Merchant AT, Pogue J, Anand SS (2007). Waist Communications Sydney Australia. The Asia Pacific Perspective:
circumference and waist-to-hip ratio as predictors of cardio- Redefining obesity and its treatment.
vascular events: meta-regression analysis of prospective studies. Yusuf S, Hawken S, Ounpuu S, Bautista L, Franzosi MG, Commerford
Eur Heart J 28, 850–856. P et al. (2005). Obesity and the risk of myocardial infarction in
Decoda Study Group. Nyamdorj R, Qiao Q, Lam TH, Tuomilehto J, 27 000 participants from 52 countries: a case-control study. Lancet
Ho SY, Pitkäniemi J et al. (2008). BMI compared with central 366, 1640–1649.

European Journal of Clinical Nutrition

You might also like