One reason why waist-to-height ratio is usually better related to chronicdisease risk and outcome than body mass index

STANLEY J. ULIJASZEK1, MACIEJ HENNEBERG2, & C.J.K. HENRY3

1Unit for Biocultural Variation and Obesity, Institute of Social and Cultural Anthropology, University of Oxford, 51 Banbury

Road, Oxford OX2 6PF, UK, 2Biological Anthropology and Comparative Anatomy Research Unit, University of Adelaide,

Adelaide SA 5005, Australia, and 3Department of Sport and Health Sciences, Oxford Brookes University, Gipsy Lane,

Oxford, UK

AbstractThe waist-to-height ratio (wtHR) has been proposed as an alternative to body mass index (BMI) as a simple anthropometricmeasure of body fatness. Both measures retain residual correlations with height, which causes them to over- or under-adjust forheight (and thus misestimate nutritional state) when relating these measures to chronic disease risk, morbidity or mortality. Thepossibility that BMI has greater misadjustment than wtHR relative to waist/heightp and weight/heightp (where p is the optimalexponent for each population and sex group) is examined here. Analysis of anthropometric data for groups in Thailand, PapuaNew Guinea and Australia shows that this is the case, especially over-adjustment. This may contribute to the weakerrelationships of chronic disease markers and outcomes with BMI than with wtHR.

Keywords: obesity, nutritional assessment, height exponents overweight

Introduction

The waist-to-height ratio (wtHR) has been proposed

as an alternative to body mass index (BMI) simple

anthropometric measure of body fatness, largely

because of the simplicity of its conceptualization and

use (Ashwell and Hsieh 2005; McCarthy and Ashwell

2006; Browning et al. 2010). It has been shown to have

similar or stronger relationships than BMI to

metabolic markers of cardiovascular disease risk

among both adults (Lin et al. 2002; Bosy-Westphal

et al. 2006; Hsieh and Muto 2006; Liu et al. 2011) and

children and adolescents (Savva et al. 2000; Hara et al.

2002; Kahn et al. 2005; Freedman et al. 2007; Weili

et al. 2007; Garnett et al. 2008). It is more strongly

associated with blood pressure (Savva et al. 2000;

Khan et al. 2008; Liu et al. 2011), and a far better

predictor of cardiovascular events and mortality

(Schneider et al. 2010) than BMI. It is also a better

predictor of type 2 diabetes risk (Hadaegh et al. 2006;

Lorenzo et al. 2007; MacKay et al. 2009; Liu et al.

2011) and metabolic syndrome (Hsieh and Muto

2006). Waist circumference (WC) alone is a better

predictor of cardiovascular disease and diabetes

outcomes than is BMI, although it can over- and

under-evaluate risk for tall and short people with

similar WC (Browning et al. 2010). Browning et al.

(2010) estimated mean values for area under receiver

operating characteristic (AUROC) curves for 147

individual AUROC analyses of wtHR, WC and BMI

in relation to chronic disease risk and physiological

markers for such risk in 31 published articles. They

found AUROC values to be higher for wtHR than for

BMI for risk of diabetes and cardiovascular disease,

and for measures of insulin resistance, hypertension,

dyslipidaemia and metabolic syndrome in both men

and women. AUROC values for WC were in all cases

lower than those for wtHR but higher than those

for BMI.

ISSN 0963-7486 print/ISSN 1465-3478 online q 2012 Informa UK, Ltd.

DOI: 10.3109/09637486.2012.734291

Correspondence: S. Ulijaszek, Unit for Biocultural Variation and Obesity, Institute of Social and Cultural Anthropology, University of Oxford,51 Banbury Road, Oxford OX2 6PE, UK. Tel: 01865 274692. Fax: 01865 274630. E-mail: stanley.ulijaszek@anthro.ox.ac.uk

International Journal of Food Sciences and Nutrition,

May 2013; 64(3): 269–273

Int J

Foo

d Sc

i Nut

r D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

26/1

4Fo

r pe

rson

al u

se o

nly.

Despite adjusting for stature, both wtHR and BMI

retain residual correlation with height (Tybor et al.

2008). This causes both measures to over- or under-

adjust for the effect of height when relating these

measures to chronic disease risk, morbidity or

mortality. The extent of such misadjustment influ-

ences the extent to which under- and over-nutrition

are estimated. It may be that one reason why wtHR

gives stronger relationships with chronic disease risk,

morbidity and mortality is that it has lower height-

related misadjustment than does BMI. Using anthro-

pometric data for three populations, in Thailand,

Papua New Guinea (PNG) and Australia, the differing

extents to which wtHR and BMI give over- and under-

adjusted estimates of nutritional state are examined.

Methods

The three samples of adults, in Thailand, PNG and

Australia, are described in detail elsewhere (Henry

et al. 2001; Ulijaszek 2003; Henneberg and Veitch

2005). In brief, the Thai sample is a composite one of

elderly people living in a residential home in

south Bangkok, and elderly living with their relatives

in north Bangkok. The non-residential subjects were

slightly taller, with greater skin folds, than their

residential counterparts, but did not differ in weight

or WC (Henry et al. 2001). The PNG sample of

adults is a rural one, based in the Purari delta of south

Coast New Guinea. The Australian sample comes

from the National Size and Shape Survey of

Australia, carried out in 2002 in the cities of Adelaide,

Brisbane, Canberra, Melbourne, Perth and Sydney.

These volunteers were overwhelmingly middle-class

Australians of European descent. Since the survey

was organized by the garment industry company,

SHARP Dummies Pty Ltd, Belair, South Australia

SA 5052, Australia and aimed at improvement in

apparel sizing standards, the majority of the volunteers

were adult females.

The extent to which wtHR and BMI are indepen-

dent of height was estimated by measuring residual

correlations with height. Within each population and

sex group, a log-log regression of WC on height, and

Table I. Sample characteristics.

Male Female Two-way analysis of variance: p (n.s.)

N Mean SD N Mean SD Population Sex Population £ sex

Thailand

Age (years) 164 71.9 7.2 273 71.8 7.5 ,0.001 n.s. n.s.

Height (cm) 164 160.1 5.7 273 150.2 6.3 ,0.001 ,0.001 ,0.01

Weight (kg) 164 53.6 8.8 273 50.7 10.8 ,0.001 ,0.001 ,0.05

Waist circumference (cm) 164 78.6 9.2 273 78.4 12.2 ,0.001 ,0.01 ,0.01

Papua New Guinea

Age (years) 142 36.5 14.2 144 35.8 14.2

Height (cm) 142 163.5 6.4 144 153.3 5.6

Weight (kg) 142 60.4 8.8 144 54.0 11.0

Waist circumference (cm) 142 77.1 5.0 144 78.2 10.2

Australia

Age (years) 52 50.6 13.9 1187 48.9 14.0

Height (cm) 52 176.0 7.3 1187 162.5 6.6

Weight (kg) 52 82.4 12.0 1187 73.6 16.2

Waist circumference (cm) 52 96.2 7.8 1187 87.4 13.3

Note: n.s., not significant.

Table II. Residual correlation between waist-to-height ratio and height, and body mass index and height.

WC/Htp Wt/Htp

CorrelationExponent

CorrelationExponent

WC/Ht-v-Ht Optimal p SE F p BMI-v-Ht Optimal p SE F p

Male

Thailand 20.10 0.65 0.25 7.0 ,0.01 0.04 2.17 0.31 48.9 ,0.001

Papua New Guinea 20.20* 0.69 0.13 28.3 ,0.001 0.02 2.15 0.26 65.8 ,0.001

Australia 20.21 0.61 0.27 5.0 ,0.05 0.01 2.03 0.40 25.5 ,0.001

Female

Thailand 20.06 0.73 0.21 11.8 ,0.001 0.15* 2.61 0.26 104.4 ,0.001

Papua New Guinea 20.08 0.73 0.28 6.9 ,0.01 0.14 2.71 0.38 49.8 ,0.001

Australia 20.17** 0.55 0.11 10.5 ,0.001 20.06 1.70 0.14 180.0 ,0.001

*p , 0.05; **p , 0.01.

S. J. Ulijaszek et al.270

Int J

Foo

d Sc

i Nut

r D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

26/1

4Fo

r pe

rson

al u

se o

nly.

weight on height, was fitted. The regression coefficient

for height from the regression model was identified

as the optimal exponent for height in the ratios

waist/heightp and weight/heightp, respectively. This is

the power that ensures that the log of the ratios is

uncorrelated with the log of height (Tybor et al. 2008).

An estimate was then obtained of the proportions of

subjects for whom wtHR was over- and under-

estimated by more or less than 10%, respectively,

relative to wtHR calculated using the optimal

exponent for height for each sex and population

group. The same procedure was carried out for

BMI. A comparison of the proportion of subjects for

whom wtHR and BMI were misadjusted by plus

and minus 10% was carried out using Chi-square

analysis. All analyses were carried out using SPSS

for PC, version 15.

Results

Mean age and anthropometric characters of the

three samples are given in Table I. Two-way analysis

of variance shows there to be great variation between

populations in age, stature, weight and WC, and

between sexes in height, weight and WC. The

populations differ significantly in age (F ¼ 677.0;

p , 0.001; two degrees of freedom), height

(F ¼ 327.0; p , 0.001; two degrees of freedom),

weight (F ¼ 230.7; p , 0.001; two degrees of freedom)

and WC (F ¼ 99.5; p , 0.001; two degrees of free-

dom). There are also significant sex differences in

height (F ¼ 628.5; p , 0.001; one degree of freedom),

weight (F ¼ 37.7; p , 0.001; one degree of freedom)

and WC (F ¼ 9.7; p , 0.01; one degree of freedom).

There is significant population by sex interaction for

WC (F ¼ 11.3; p , 0.001; two degrees of freedom),

stature (F ¼ 5.8; p , 0.01; two degrees of freedom)

and weight (F ¼ 3.2; p , 0.05; two degrees of

freedom), due mainly to large male–female differences

in WC and stature in the Australian population, and

weight differences between the sexes which are largest

for the Australian population, and smallest for the

Thai population. Table II gives the residual corre-

lations between wtHR and height, and between BMI

and height, as well as showing the optimal exponents

for both WC and weight in their relationship with

height. Correlations with height are small for both

wtHR and BMI. The optimal exponent for WC in

relation to height to attain the best height indepen-

dence ranges between 0.55 and 0.73, while the optimal

exponent for weight in relation to height to attain the

best height independence ranges between 1.70 and

2.71. There is no systematic difference in optimal

exponent for either WC or weight between the sexes,

although the optimal exponents for both Australian

males and females are lower than the other two

populations.

Table III gives the proportion of subjects for whom

wtHR and BMI give either under- or over-adjustmentTable

III.

Pro

port

ion

ofsu

bje

cts

for

whom

wais

t-to

-hei

ghtra

tio

an

dB

MI

giv

eei

ther

un

der

-or

over

-ad

just

men

tofn

utr

itio

nalst

ate

of10%

or

more

,re

lati

ve

tow

ais

t/h

eigh

tpan

dw

eight/

hei

gh

tp,w

her

ep

isth

e

opti

mal

expon

ent

for

each

popu

lati

on

an

dse

xgro

up

(n.s

.,n

ot

sign

ifica

nt)

.

Th

ailan

dP

ap

ua

New

Gu

inea

Au

stra

lia

wtH

RB

MI

wtH

RB

MI

wtH

RB

MI

N%

N%

Ch

i-sq

uare

,

wtH

R-v

-BM

I,p

N%

N%

Ch

i-sq

uare

,

wtH

R-v

-BM

I,p

N%

N%

Ch

i-sq

uare

,

wtH

R-v

-BM

I,p

Male

s

.10%

over

-ad

just

men

t29

17.7

44

26.8

,0.0

16

4.2

26

18.3

,0.0

01

11.9

13

25.0

,0.0

01

Est

imate

wit

hin

^10%

92

56.1

83

50.6

n.s

.126

88.8

88

62.0

,0.0

01

42

80.8

27

51.9

,0.0

5

.10%

un

der

-ad

just

men

t43

26.2

37

22.6

n.s

.10

7.0

28

19.7

,0.0

01

917.3

12

23.1

n.s

.

Tota

l164

100

164

100

,0.0

5142

100

142

100

,0.0

01

52

100

52

100

,0.0

01

Fem

ale

s

.10%

over

-ad

just

men

t59

21.6

79

28.9

,0.0

122

15.3

41

28.5

,0.0

01

261

22.0

428

36.1

,0.0

01

Est

imate

wit

hin

^10%

132

48.4

129

47.3

n.s

.88

61.1

72

50.0

n.s

.561

47.3

437

36.8

,0.0

01

.10%

un

der

-ad

just

men

t82

30.0

65

23.8

,0.0

534

23.6

31

21.5

n.s

.365

30.7

322

27.1

,0.0

1

Tota

l273

100

273

100

,0.0

1144

100

144

100

,0.0

11187

100

1187

100

,0.0

01

Why waist-to-height ratio is better 271

Int J

Foo

d Sc

i Nut

r D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

26/1

4Fo

r pe

rson

al u

se o

nly.

of 10% or more, relative to waist/heightp and

weight/heightp, where p is the optimal exponent for

each population and sex group, as given in Table II.

This shows both wtHR and BMI to give high degrees

of both over- and under-adjustment of nutritional

state relative to the use of the optimal exponent,

in both sexes of all three populations. The BMI over-

adjusts for height significantly more than does wtHR

among both sexes of all three populations, and under-

adjusts for height more among PNG males, and

less among Thai and Australian females.

Discussion

The wtHR is easier to measure and calculate than

BMI, and is a better identifier of subjects at

increased risk of cardiovascular disease, hyperten-

sion, diabetes and the metabolic syndrome. Both

measures adjust for stature, both far from perfectly.

This misadjustment is a source of error that may

weaken relationships among risk markers for chronic

disease and nutritional state as estimated by these

measures. The BMI gives significantly more mis-

adjustment, especially over-adjustment, than wtHR,

and this may well contribute to its weaker

relationships with diabetes, cardiovascular disease,

insulin resistance, dyslipidaemia and metabolic

syndrome (Browning et al. 2010). The extent of

this misadjustment may vary from population to

population, and a larger study of adults is needed to

estimate it more precisely. Misadjustment is also

expected when these measures are used to predict

chronic disease risk among children and adolescents,

but needs to be demonstrated. A further problem

with BMI may be that exponential functions may

give better measures of relative weight for height

than allometric functions (Henneberg et al. 1989).

As a linear function, wtHR does not have that

problem, and this may be another reason why wtHR

may be a more appropriate measure of body size

than BMI, particularly in the assessment of over-

weight and obesity.

Conclusions

Both BMI and wtHR are simple stature-corrected

proxies for nutritional state used in the assess-

ments of health risk associated with overweight and

obesity. This correction for stature is imperfect,

and more so for BMI than for wtHR. Misadjustment

for stature is likely to be one reason for the lower

associations between chronic disease and its markers

and BMI than between chronic disease and its

markers and wtHR.

Acknowledgements

We thank Ms Daisy Veitch of SHARP Dummies

Pty Ltd.

Declaration of interest: The authors report no

conflicts of interest. The authors alone are responsible

for the content and writing of the paper.

References

Ashwell M, Hsieh SD. 2005. Six reasons why the waist-to-height

ratio is a rapid and effective global indicator for health risks

of obesity and how its use could simplify the international

public health message on obesity. Int J Food Sci Nutr 56:

303–307.

Bosy-Westphal A, Geisler C, Onur S, Korth O, Selberg O,

Schrezenmeir J, Muller MJ. 2006. Value of body fat mass vs

anthropometric obesity indices in the assessment of metabolic

risk factors. Int J Obes 30:475–483.

Browning LM, Hsieh SD, Ashwell M. 2010. A systematic review of

waist-to-height ratio as a screening tool for the prediction of

cardiovascular disease and diabetes: 0.5 could be a suitable

global boundary value. Nutr Res Rev 23:247–269.

Freedman DS, Kahn HS, Mei Z, Grummer-Strawn LM, Dietz WH,

Srinivasan SR, Berenson GS. 2007. Relation of body mass index

and waist-to-height ratio to cardiovascular disease risk factors in

children and adolescents: the Bogalusa Heart Study. Am J Clin

Nutr 86:33–40.

Garnett SP, Baur LA, Cowell CT. 2008. Waist-to-height ratio: a

simple option for determining excess central adiposity in young

people. Int J Obes 32:1028–1030.

Hadaegh F, Zabetian A, Harati H, Azizi F. 2006. Waist/height ratio

as a better predictor of type 2 diabetes compared to body mass

index in Tehranian adult men – a 3.6-year prospective study.

Exp Clin Endocrinol Diabetes 114:310–315.

Hara M, Saitou E, Iwata F, Okada T, Harada K. 2002. Waist-to-

height ratio is the best predictor of cardiovascular disease risk

factors in Japanese schoolchildren. Atheroscler Thromb 9:

127–132.

Henneberg M, Veitch D. 2005. Is obesity as measured by body mass

index, and waist circumference in Adult Australian Women 2002

just a result of the lifestyle? J Hum Ecol 13:85–89.

Henneberg M, Hugg J, Townsend EJ. 1989. Body weight/height

relationship: exponential solution. Am J Hum Biol 1:483–491.

Henry CJK, Varakamin C, Webster-Gandy J, Ulijaszek SJ. 2001.

Anthropometry of two contrasting populations of Thai elderly

living in a rural setting. Arch Gerontol Geriatr 33:255–263.

Hsieh SD, Muto T. 2006. Metabolic syndrome in Japanese men and

women with special reference to the anthropometric criteria for

the assessment of obesity: proposal to use the waist-to-height

ratio. Prev Med 42:135–139.

Kahn HS, Imperatore G, Cheng YJ. 2005. A population-based

comparison of BMI percentiles and waist-to-height ratio

for identifying cardiovascular risk in youth. J Pediatr 146:

482–488.

Khan A, Haq FU, Pervez MB, Saleheen D, Frossard PM, Ishaq M,

et al. 2008. Anthropometric correlates of blood pressure in

normotensive Pakistani subjects. Int J Cardiol 124:259–262.

Lin WY, Lee LT, Chen CY, Lo H, Hsia HH, Liu IL, et al. 2002.

Optimal cut-off values for obesity: using simple anthropometric

indices to predict cardiovascular risk factors in Taiwan. Int J Obes

Rel Metabol Disoders 26:1232–1238.

Liu Y, Tong G, Tong W, Lu L, Qin X. 2011. Can body mass index,

wasit circumference, waist-hip ratio and waist-height ratio

predict the presence of multiple metabolic risk factors in Chinese

subjects? BMC Public Health 11:35.

Lorenzo C, Serrano-Rios M, Martinez-Larrad MT, Gonzalez-

Villalpando C, Williams K, Gabriel R, et al. 2007. Which obesity

index best explains prevalence differences in type 2 diabetes

mellitus? Obesity 15:1294–1301.

MacKay MF, Haffner SM, Wagenknecht LE, D’Agostino RB,

Hanley AJG. 2009. Prediction of type 2 diabetes using alternate

S. J. Ulijaszek et al.272

Int J

Foo

d Sc

i Nut

r D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

26/1

4Fo

r pe

rson

al u

se o

nly.

anthropometric measures in a multi-ethnic cohort. Diabetes

Care 32:956–958.

McCarthy HD, Ashwell M. 2006. A study of central fatness using

waist-to-height ratios in UK children and adolescents over two

decades supports the simple message – ‘keep your waist cir-

cumference to less than half your height’. Int J Obes 30:988–992.

Savva SC, Tornaritis M, Savva ME, Kourides Y, Panagi A,

Silikiotou N, et al. 2000. Waist circumference and waist-to-height

ratio are better predictors of cardiovascular disease risk factors

in children than body mass index. Int J Obes 24:1453–1458.

Schneider HJ, Friedrich N, Klotsche J, Pieper L, Nauck M, John U,

et al. 2010. The predictive value of different measures of obesity

for incident cardiovascular events and mortality. J Clin

Endocrinol Metab 95:1777–1785.

Tybor DJ, Lichtenstein AH, Dallal GE, Must A. 2008. Waist-

to-height ratio is correlated with height in US children

and adolescents ages 2–18 years. Int J Pediatr Obes 3:

148–151.

Ulijaszek SJ. 2003. Socioeconomic factors associated with physique

of adults of the Purari delta of the Gulf Province, Papua New

Guinea. Ann Hum Biol 30:316–328.

Weili Y, He B, Yao H, Dai J, Cui J, Ge D, et al. 2007. Waist-to-height

ratio is an accurate and easier index for evaluating obesity in

children and adolescents. Obesity 15:748–752.

Why waist-to-height ratio is better 273

Int J

Foo

d Sc

i Nut

r D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

26/1

4Fo

r pe

rson

al u

se o

nly.