Tiffany Gill 1, Catherine Chittleborough 1, Anne Taylor 1, Section: International comparison of health determinants

2 3 4
Richard Ruffin , David Wilson , Patrick Phillips

1
Centre for Population Studies in Epidemiology, Department of Human Services, Adelaide
2
Department of Medicine, The Queen Elizabeth Hospital, Woodville South
3
Department of Medicine, The University of Adelaide
4
Endocrinology, The Queen Elizabeth Hospital, Woodville South

Body mass index, waist hip ratio, and waist circumference:

which measure to classify obesity?

2001; Ho et al. 2001; Lev-Ran 2001; McCarthy et al. 2001;

Summary IARC 2001; Hu et al. 2000). Laboratory measures of obesity,
such as body potassium, body density and total body water
Objectives: To determine the proportion of a representative are expensive in terms of time and equipment costs and are
population sample of adults in South Australia who have a impractical for use in large-scale population studies (Reilly
body mass index (a measure of overall obesity) classified as nor- et al. 2000; Bandini & Dietz 1987). As a result, anthropo-
mal or underweight, but who also have a waist circumference metric measures and associated indices have been used in an
or waist hip ratio (measures of central obesity) that indicates attempt to identify those with a high level of obesity and who
obesity. may be at risk of developing chronic health conditions.
Methods: A representative population sample of adults aged Body mass index (BMI), based on self-reported or clinically
18 years and over living in the north west region of Adelaide assessed height and weight, is an accepted, inexpensive mea-
(n = 2 523) were recruited to the study. Clinical measures of sure to determine overweight or obesity used in population
height, weight, waist and hip circumference were obtained studies due to the general ease and accuracy of the mea-
and used to determine body mass index, waist hip ratio and surement required in order to calculate the index (Malina &
waist circumference. Katzmarzyk 1999). BMI is determined by dividing weight (in
Results: Among women with a normal body mass index, 19.0 % kilograms) by height (in metres2) and is regarded as a mea-
had a high waist circumference (≥ 80 cm) and 8.5 % had a high sure of the total body fat (Doll et al. 2002; Lev-Ran 2001),
waist hip ratio (> 0.85). Among males with a normal body mass that is, a measure of overall obesity. A high BMI has been
index, 3.4 % had a high waist circumference (≥ 95 cm) and 0.1 % identified as a risk factor for development of conditions such
had a high waist hip ratio (>1.0). as diabetes and cardiovascular disease (Hu et al. 2000; WHO
Conclusions: Body mass index, waist hip ratio and waist cir- 1998; Australian Centre for Diabetes Strategies 2000).
cumference all have a role in the identification of those who An android or centralised pattern of fat distribution, where
are obese or overweight. excess body fat is distributed in the abdominal region rather
than on the hips and thighs, has also been demonstrated to
be an important consideration in chronic disease incidence
Keywords: Body mass index – Height – Weight – Anthropometry –
(Despres et al. 2001; Ho et al. 2001, Lev-Ran 2001; Aus-
Abdominal obesity – Central obesity.
tralian Centre for Diabetes Strategies 2000; WHO 1998;
Lean et al. 1995). Waist circumference (WC) and hip cir-
The prevalence of obesity is increasing in many developed cumference (HC) are the two basic anthropometric mea-
and developing countries (Doll et al. 2002; Ho et al. 2001; surements taken and WC is divided by HC in order to create
Lev-Ran 2001). Obesity has been linked with the incidence the waist hip ratio (WHR). Both WC and WHR provide
of chronic conditions such as diabetes, cancer, cardiovascu- an indication of abdominal obesity (Doll et al. 2002; Ho
lar disease, hypertension, hyperglycemia, and hypercholes- et al. 2001; Assayama et al. 2000; Lev-Ran 2001; Lean et al.
terolemia (Doll et al. 2002; Defay et al. 2001; Despres et al. 1995).
Soz.- Präventivmed. 48 ( 2003) 191 – 200
0303-8408/03/030191–10
DOI 10.1007/s00038-003-2055-1
© Birkhäuser Verlag, Basel, 2003
192 Section: International comparison of health determinants Gill T, Chittleborough C, Taylor A, et al.
Body mass index, waist hip ratio, and waist circumference

In order to determine whether populations are at risk of be sufficiently sensitive as a risk factor, as discrepancies can
chronic disease, cutoff values have been determined for occur in defining obese populations, depending on the mea-
BMI, WC and WHR. These values provide an indication of surement of central or overall obesity used (Lev-Ran 2001;
the proportion of the population “at risk” of chronic dis- Roubenoff et al. 1995). Booth et al. (2000) examined the re-
eases, although even those that gain weight within the range lationship between BMI and WC for males and females and
considered “normal” may be at greater risk of diseases such found that the prevalence of overweight based on self-re-
as cancer (IARC 2001). The cut-off values for BMI are ported BMI and WC combined was significantly greater
25 kg/m2, which indicates an overweight adult, and 30 kg/m2, than the prevalence based on self-reported BMI alone.
which identifies an obese adult (Cole et al. 2000; WHO This paper examines measurements of BMI, WC and WHR
1998). taken on a population sample to determine the level of over-
Several cut-offs have been proposed for WC and WHR, in- all and central obesity using different methods of classifying
dicators of abdominal fat accumulation, based on their asso- obesity. The WC and WHR of those subjects with a normal
ciation with relative risk of disease (Molarius & Seidell or underweight BMI were determined in order to assess
1998). Studies conducted by Han et al. (1995) and Lean et al. whether these subjects would be classified as obese using dif-
(1995) suggested that men with a WC of 94 cm or greater and ferent criteria. Demographic characteristics of those classi-
women with a WC of 80 cm or greater should not gain fur- fied with a normal or underweight BMI but high central obe-
ther weight. A WC of 102 cm or greater for men and 88 cm sity are also examined.
or greater for women, was the level suggested for weight re-
duction to occur (Han et al. 1995; Lean et al. 1995). These
cut-off values were also used by Booth et al. (2000). Adop- Methods
tion of these recommended “action levels” was suggested by
the Australian Centre for Diabetes Strategies (2000) with an Sample selection
adjustment for men to a WC of 95 cm or greater as the end- The sample described in this paper was obtained from the
point for not gaining more weight, and the reduction of North West Adelaide Health Study, a collaboration between
weight limits being 100 cm or greater for men and 90 cm or the North Western Adelaide Health Service (The Queen
greater for women. This paper defines a high WHR as 1.0 for Elizabeth Hospital and Lyell McEwin Health Service cam-
males and 0.85 for females (Australian Centre for Diabetes puses), the South Australian Department of Human Ser-
Strategies 2000). These cut-offs, however, vary according to vices, The University of Adelaide, and the University of
the population groups being described and care is required South Australia. This study is one of the first of its kind in
when examining associations for populations other than Australia to provide a comprehensive health assessment of
Caucasians (Lev-Ran 2001; Australian Centre for Diabetes specified chronic health conditions in the adult population
Strategies 2000; Booth et al. 2000; Molarius et al. 1999; Mo- of a particular community (Centre for Population Studies in
larius & Seidell 1998; Han et al. 1995; Lean et al. 1995). Epidemiology 2002a; 2002b). The north west region of Ade-
The risk of chronic disease has been shown to vary accord- laide was considered a priority study area because it is iden-
ing to the measure of obesity used. Strong positive associa- tified as having overall greater relative social disadvantage
tions have been demonstrated between BMI, hypertension, compared with many other areas of South Australia and pre-
cardiovascular disease and diabetes (Despres et al. 2001; vious evidence from population surveys (Wilson et al. 1992)
Lev-Ran 2001; Hu et al. 2000; WHO 1998). Defay et al. and studies indicate higher levels of chronic disease and risk
(2001) found that in older males the prevalence of diabetes factors in the region (Pilotto et al. 1999).
was associated with total body fat whereas in older females All households in the north western area of Adelaide with a
abdominal fat as measured by WHR was associated with di- telephone connected and a number listed in the Electronic
abetes. Ho et al. (2001) also demonstrated an effect of gen- White Pages (EWP) were eligible for selection in the study.
der on the association of BMI, WC and WHR and metabolic A household was randomly selected from the EWP and an
syndrome, with BMI and WC useful in assessing cardiovas- introductory letter and an information brochure were sent
cular risk factors in males and WC and WHR the most use- to the householder. Within each household, the person aged
ful for women. Other studies have also indicated an associa- 18 years or over who was last to have a birthday was selected
tion of WC with metabolic functions and coronary heart dis- and interviewed using Computer-Assisted Telephone Inter-
ease and increased mortality (Defay et al. 2001; Despres et view (CATI) technology and invited to attend a clinic as-
al. 2001; McCarthy et al. 2001; Visscher et al. 2001; Lean et sessment. Up to 10 call-backs were made to each selected
al. 1995). There is also evidence to suggest that BMI may not household to obtain an interview. There was no replacement

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
Gill T, Chittleborough C, Taylor A, et al. Section: International comparison of health determinants 193
Body mass index, waist hip ratio, and waist circumference

for refusal or non-response. During the recruitment process, males or ≥ 80 cm for females was defined as the level at
appointments were made to call back pregnant women after which no further weight should be gained. A WC of ≥ 100 cm
their expected delivery date. for males and ≥ 90 cm for females was defined as the level at
which weight reduction should be recommended (Han et al.
Data collection 1995; Lean et al. 1995).
During the telephone recruitment interview, information
was collected on contact details, demographic information, Statistical analyses
self-reported health conditions, mental health, smoking sta- The data were weighted to the Australian Bureau of Statis-
tus and where respondents refused to take part in the study, tics 1999 Estimated Residential Population (Australian
the reasons behind the refusal. Demographic questions in- Bureau of Statistics 2000) by region, age group, sex, and
cluded age, sex, area of residence, highest education level probability of selection in the household, to ensure that the
obtained, gross household income, country of birth, marital sample was representative of the North West region of
status, and work status. If participants agreed to attend for a Adelaide. The data were analysed using SPSS version 10.0
clinic assessment, an appointment was made at either The (Statistical Package for the Social Sciences 1999) and Epi
Queen Elizabeth Hospital or the Lyell McEwin Health Ser- Info 6.0 (Dean et al. 1994). Chi-square tests were perform-
vice, two teaching hospitals located in the western and ed to compare the prevalence of overweight and obesity
northern suburbs of Adelaide, South Australia. Participants for each measure between males and females. The propor-
were sent an information folder that contained information tion of males and females with normal or underweight BMI
about the location of the clinic, confirmation of their ap- who also had a high WHR or WC was determined. Univari-
pointment and a further questionnaire that they completed ate analyses were conducted to determine odds ratios for de-
and returned when they attended their appointment. This mographic variables associated with a normal or under-
questionnaire also examined issues such as quality of life, weight BMI but a high WHR or WC among females. Males
mental health, health service use, risk factor prevalence and were not included in these analyses because only a small
health conditions in more detail. number were classified with a normal or underweight BMI
Measurements taken within the clinic setting included blood but a high WHR or WC. Supplementary tables in the Ap-
pressure and spirometry. A fasting blood sample was taken pendix report the percentiles P10, P25, P50, P75, P90, as well
to determine cholesterol and glucose levels and skin tests as the mean and standard deviation (SD) for the continuous
were performed to test sensitivity to a variety of irritants. variables of BMI, WHR, and WC for males and females by
The four measurements taken to determine the level of obe- age group.
sity were height, weight, waist circumference and hip cir-
cumference. These measurements were taken by assistants
who had undergone rigorous training prior to the com- Results
mencement of the clinic sessions. Of the total eligible sample (n = 4 951), 74% (n = 3 650) took
Height was measured to the nearest 0.5 cm using a sta- part in the initial telephone interview and 51% (n = 2 523)
diometer, and weight to the nearest 0.1 kg in light clothing attended the clinic. The response rate for attendance at the
and without shoes using standard digital scales. BMI was cal- clinic among those who were interviewed was 69%.
culated as weight (kg)/height (m)2. Overweight was defined The distributions of BMI, WHR, and WC for males and fe-
as BMI > 25 kg/m2 and obesity as BMI > 30 kg/m2 (National males by age group are shown in the Appendix. The overall
Heart Foundation 1989). prevalence of overweight, as measured by BMI, was 35.2%
WC was measured to the nearest 0.1 cm using an inelastic (95% CI: 33.3–37.1) and the prevalence of obesity was
tape maintained in a horizontal plane, with the subject 28.5% (95% CI: 26.8–30.4). Overall, 16.7% (95% CI: 15.3–
standing comfortably with weight distributed evenly on both 18.3) were classified as having a high WHR. According to
feet. The measurement was taken at the level of the narrow- the WC measure, 57.4% (95% CI: 55.4–59.3) were at the
est part of the waist. The mean of three measurements was level where they should be advised to gain no further weight
calculated. Hip circumference was also measured using an and 36.7% (95% CI: 34.8–38.6) were at the level where they
inelastic tape, at the level of the maximum posterior exten- should be advised to lose weight. Table 1 shows the preva-
sion of the buttocks. Three measurements were taken and lence of overweight as measured by BMI, WHR and WC, by
the average of the three was calculated. A high WHR was sex. Males were statistically significantly more likely than fe-
defined as > 1.0 for males and > 0.85 for females (Australian males to be overweight or obese according to their BMI. Fe-
Centre for Diabetes Strategies 2000). A WC of ≥ 95 cm for males, however, were statistically significantly more likely to

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
194 Section: International comparison of health determinants Gill T, Chittleborough C, Taylor A, et al.
Body mass index, waist hip ratio, and waist circumference

Table 1 Prevalence of overweight and obesity according to BMI, WHR, and WC

Males Females Overall

n % n % n %

BMI (> 25, overweight/obese)

no 377 30.5 ∨ 538 41.8 ∧ 915 36.3
yes 859 69.5 ∧ 749 58.2 ∨ 1608 63.7
WHR (obese) (> 1.0 men, > 0.85 women)
no 1145 92.7 ∧ 956 74.2 ∨ 2101 83.3
yes 90 7.3 ∨ 332 25.8 ∧ 422 16.7
WC (lose weight)
(≥ 100 cm men, ≥ 90 cm women)
no 788 63.8 808 62.8 1597 63.3
yes 447 36.2 479 37.2 926 36.7
WC (not gain weight)
(≥ 95 cm men, ≥ 80 cm women)
no 567 45.9 ∧ 508 39.5 ∨ 1075 42.6
yes 668 54.1 ∨ 779 60.5 ∧ 1448 57.4

∧ ∨ Statistically significantly higher or lower than comparison gender group.

have a high WHR or a WC where they should be advised to should be advised to gain no further weight, 12.6% of par-
gain no further weight. ticipants (3.4% of males, 19.0% of females) with normal or
The proportion of participants with normal or underweight underweight BMI had a WC at this level. Among females
BMI, by age and sex, who had a high WHR or WC are aged 60 years and over with normal or underweight BMI,
shown in Table 2. Overall, 5.0 % of people with normal or 43.1% had a WC at this level.
underweight BMI had a high WHR. Only 0.1 % of males Of those participants with a high WHR, 10.9% (n = 47) were
with normal or underweight BMI had a high WHR. Among classified as normal or underweight according to their BMI
females with normal or underweight BMI, however, 8.5% (9.9% were normal and 0.9% were underweight). Of partici-
overall had a high WHR, and 26.3 % aged 60 years or over pants with a WC at the level where they should be advised to
had a high WHR. gain no further weight, 7.8% (n = 114) were classified as nor-
Among participants who were normal or underweight ac- mal or underweight according to their BMI (7.6% were nor-
cording to their BMI, 1.1 % (0.6 % of males, 1.5 % of fe- mal and 0.2% were underweight). Of participants with a WC
males) had a WC at the level where they should be advised at the level where they should be advised to lose weight, 0.9%
to lose weight. In terms of a WC at the level where they (n = 9) were classified as normal according to their BMI.

Table 2 Proportion (%) of participants with normal or underweight BMI, by sex and age group, who have high WHR and/or WC

BMI ≤ 25 kg/m2 WHR WC

Males n ≤ 1.0 > 1.0 < 100 cm ≥ 100 cm a < 95 cm ≥ 95 cm b

Age group (years)

18 – 39 247 100.0 – 100.0 – 99.3 0.7
40 – 59 75 99.6 0.4 99.6 0.4 93.1 6.9
60+ 55 100.0 – 96.3 3.7 89.2 10.8
All ages 377 99.9 0.1 99.4 0.6 96.6 3.4

Females n ≥ 0.85 > 0.85 < 90 cm ≥ 90 cm a < 80 cm ≥ 80 cm b

Age group (years)

18 – 39 324 96.1 3.9 99.0 1.0 89.2 10.8
40 – 59 136 91.0 9.0 99.1 0.9 75.3 24.7
60+ 78 73.7 26.3 95.5 4.5 56.9 43.1
All ages 538 91.5 8.5 98.5 1.5 81.0 19.0
Overall 915 95.0 5.0 98.9 1.1 87.4 12.6

a
Action level at which weight should be reduced.
b
Action level at which no further weight should be gained.

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
Gill T, Chittleborough C, Taylor A, et al. Section: International comparison of health determinants 195
Body mass index, waist hip ratio, and waist circumference

Among males with a high WHR, 0.3 % had normal or un- Discussion
derweight BMI. In the case of WC, 1.9 % of males with a Obesity continues to be a major health risk factor in today’s
high WC (≥ 95 cm) had a normal or underweight BMI. society and can be considered a disease in its own right
Among females, with a high WHR, 13.7 % had a normal or (WHO 1998). Difficulties remain, however, as to how best to
underweight BMI. Table 3 describes the demographic char- measure it on a large scale or population basis. Measure-
acteristics associated with normal or underweight BMI ments such as BMI, WC and WHR have all been used in an
among females with a high WHR. That is, the characteristics attempt to define obesity and overweight.
of females who would be missed, or not classified as being Cut-off points are used for anthropometric measures to de-
overweight or obese, if only the BMI and not the WHR mea- scribe associations between obesity and relative risk of disease
sure was used. Among females with a high WHR, those with (Molarius & Seidell 1998). Despite the growing body of evi-
a normal or underweight BMI were significantly more likely dence supporting specific cut-off points, their selection is
than those with overweight or obese BMI to have a bachelor largely arbitrary (Molarius & Seidell 1998). Cut-off points con-
degree or higher level of education, have been born in Asia tinue to be used because of their importance in setting public
or an “other” country and be widowed, and less likely to health recommendations and comparing populations (Molar-
be aged 40 to 59 years and live in the northern suburbs of ius & Seidell 1998). It must be recognised, however, that BMI,
Adelaide. WHR, and WC are continuous variables and the risk of dis-
Among females with a high WC (≥ 80 cm), 13.0 % had a nor- ease increases, often linearly, with increasing obesity.
mal or underweight BMI. Table 4 shows the demographic The results described in this paper suggest that different “fat-
characteristics associated with normal or underweight BMI ness” criteria are measured when determining obesity using
among females who had a high WC (≥ 80 cm). Among fe- BMI, WHR or WC measurements. According to BMI mea-
males with a high WC, those with a normal or underweight surements, 63.7% of participants were defined as overweight
BMI were significantly more likely than those with an over- or obese. In contrast, 57.4% of participants would be advised
weight or obese BMI to have a bachelor degree or higher to not gain further weight and 36.7% to lose weight if their
level of education and be widowed, and less likely to be born WC measurement was used as the indicator of obesity. When
in the northern suburbs of Adelaide. considering WHR, 16.7% of participants had a high WHR. In

Table 3 Univariate odds ratios for

demographic variables associated Variable n % OR (95 % CI OR) p value
with normal or underweight BMI
among females with a high WHR Age group (years)
18 to 39 13/61 20.6 1.00
40 to 59 12/129 9.5 0.41 (0.17 – 0.95) 0.04
60+ 21/141 14.6 0.66 (0.30 – 1.43) 0.3
Area of residence
western suburbs 23/118 19.7 1.00
northern suburbs 22/213 10.4 0.47 (0.25 – 0.89) 0.02
Highest education level obtained
secondary 22/199 11.0 1.00
trade/Apprenticeship/Cert/Diploma 14/100 14.1 1.32 (0.65 – 2.71) 0.4
bachelor degree or higher 6/17 38.9 5.14 (1.74 – 15.17) 0.003
Gross household income
up to $20 000 23/129 17.5 1.00
$20 001–40 000 7/76 9.1 0.47 (0.19 – 1.17) 0.1
$40 001–60 000 4/50 7.6 0.39 (0.12 – 1.21) 0.1
more than $60 000 9/42 20.0 1.18 (0.49 – 2.85) 0.7
Country of birth
Australia 27/202 13.2 1.00
UK or Ireland 11/80 13.2 0.99 (0.46 – 2.14) 0.9
Europe, USSR, Baltic States 3/33 9.5 0.69 (0.20 – 2.37) 0.6
Asia, Other 5/11 45.8 5.54 (1.59 – 19.35) 0.007
Marital status
married or living with partner 25/212 12.0 1.00
separated/Divorced 2/33 5.0 0.39 (0.08 – 1.94) 0.2
widowed 14/51 27.0 2.71 (1.29 – 5.69) 0.008
never married 5/32 14.0 1.19 (0.41 – 3.49) 0.7
Work status
employed 15/103 14.1 1.00
unemployed/Home duties/
retired/Student 29/221 13.0 0.91 (0.46 – 1.79) 0.8

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
196 Section: International comparison of health determinants Gill T, Chittleborough C, Taylor A, et al.
Body mass index, waist hip ratio, and waist circumference

Table 4 Univariate odds ratios for

demographic variables associated Variable n % OR (95 % CI OR) p value
with normal or underweight BMI
among females with a high WC Age group (years)
(≥ 80 cm) 18 to 39 33/223 15.0 1.00
40 to 59 34/284 11.8 0.76 (0.45 – 1.27) 0.3
60+ 34/271 12.5 0.81 (0.48 – 1.35) 0.4
Area of residence
western suburbs 53/297 17.9 1.00
northern suburbs 48/480 9.9 0.50 (0.33 – 0.77) 0.001
Highest education level obtained
secondary 47/435 10.8 1.00
trade/Apprenticeship/Cert/Diploma 35/251 14.0 1.34 (0.84 – 2.14) 0.2
bachelor degree or higher 15/59 25.7 2.85 (1.48 – 5.49) 0.002
Gross household income
up to $20 000 40/268 15.0 1.00
$20 001–40 000 17/180 9.5 0.59 (0.32 – 1.08) 0.09
$40 001–60 000 16/145 11.3 0.72 (0.39 – 1.33) 0.3
more than $60 000 23/126 18.4 1.28 (0.73 – 2.24) 0.4
Country of birth
Australia 73/514 14.1 1.00
UK or Ireland 18/156 11.3 0.78 (0.45 – 1.35) 0.4
Europe, USSR, Baltic States 5/75 7.3 0.48 (0.19 – 1.18) 0.1
Asia, Other 4/21 20.8 1.60 (0.53 – 4.77) 0.4
Marital status
married or living with partner 63/521 12.1 1.00
separated/Divorced 10/70 14.3 1.21 (0.59 – 2.49) 0.6
widowed 19/92 20.7 1.89 (1.07 – 3.34) 0.03
never married 8/82 9.5 0.76 (0.35 – 1.67) 0.5
Work status
employed 46/305 15.0 1.00
unemployed/Home duties/
retired/Student 51/458 11.2 0.71 (0.46 – 1.09) 0.1

addition, 1.8% of participants had a BMI of 25 kg/m2 or less measures of obesity among older women. The significant as-
but their WHR was over the cutoff point of 1.0. sociation of Asian or “other” nationalities in the case of
Further support for the fact that BMI and WC or WHR WHR may indicate the variation in body fat distribution
measure different premises is provided by the fact that while among different ethnic groups (Lev-Ran 2001; Molarius &
some subjects were classified with a low or normal BMI, the Seidell 1998; Deurenberg et al. 1999).
WC measurement was within the action zone areas (either The results of this study indicate that while BMI is widely
no more weight should be gained, or weight should be lost). used to describe overweight and obesity, it may not be sensi-
This supports the view that there may be not be recognition tive enough to detect all those who are overweight or obese
of those at risk of chronic disease if BMI alone is used (Lev- and ultimately those who may be at risk of chronic diseases.
Ran 2001; Roubenoff et al. 1995). WC measures central adiposity and this measurement also
Participants with a low or normal BMI but a high WHR or classifies those who should not gain further weight or who
WC were significantly more likely to be female. Investiga- should lose weight. Females in particular appear more likely
tion of the demographic characteristics of females with a to be misclassified if BMI alone is used as a measure of obe-
high WHR or WC showed those with a normal or under- sity. At this stage the use of both measurements is warranted
weight BMI were more likely to have a bachelor degree or as different types of obesity are being assessed. Both central
higher level education, be widowed and living in the western adiposity and total body fat appear to have a role in the iden-
suburbs. Females have different body shapes to males and tification of those who are obese.
consequently a different distribution of fat (Perissinotto et The North West Adelaide Health Study has been designed
al. 2002; Ho et al. 2001; Lev-Ran 2001). Although age itself as a prospective cohort study. Future analyses of this sample
was not a significant predictor of normal or underweight will therefore have the potential to determine the ability of
BMI among women with a high WC, being widowed is an in- each indicator of obesity to predict morbidity and mortality.
dicator of older age, and provides an indication that BMI The fact remains, however, that regardless of which measure
and WHR or WC may change in older females as a result of is used to define obesity, the same policy development and
ageing and redistribution of fat (Perissinotto et al. 2002; Lev- health promotion initiatives are critical to reduce this public
Ran 2001). WC and WHR may therefore be more sensitive health epidemic.
Soz.- Präventivmed. 48 ( 2003) 191 – 200
© Birkhäuser Verlag, Basel, 2003
Gill T, Chittleborough C, Taylor A, et al. Section: International comparison of health determinants 197
Body mass index, waist hip ratio, and waist circumference

Zusammenfassung Résumé

Body-Mass-Index, Taille-Hüft-Quotient und Taillenumfang: Indice de masse pondérale, rapport taille/hanche et circon-
welcher ist das richtige Mass für die Bestimmung von Über- férence de la taille: quelle est la bonne mesure de l’obésité?
gewicht? Objectifs: Déterminer la proportion d’un échantillon représen-
Zielsetzungen: Die Bestimmung des Anteils an Erwachsenen in tatif de la population adulte du sud de l’Australie ayant un in-
einer repräsentativen Bevölkerungsstichprobe in Südaustra- dice de masse corporelle (IMC, une mesure de l’obésité
lien, die einen als normal oder untergewichtig geltenden Body- générale) classé comme normal ou pas, mais qui ont aussi une
Mass-Index aufweisen (BMI, ein Mass zur Bestimmung von ge- circonférence de la taille ou un rapport taille/hanche (qui sont
nerellem Übergewicht), aber gleichzeitig einen für Adipositas des mesures d’obésité centrale) compatibles avec une obésité.
typischen Taillenumfang oder Taille-Hüft-Quotient (ein Mass Méthodes: Un échantillon représentatif de la population
der zentralen Adipositas) aufweisen. adulte âgé de 18 ans et plus vivant dans la région Nord-Ouest
Methoden: Eine repräsentative Bevölkerungsstichprobe von d’Adélaide (n = 2523) a été recruté dans cette étude. Les
Erwachsenen ab 18 Jahren, die im Nordwesten von Adelaide mesures cliniques de la taille, du poids, la circonférence de la
(n = 2 523) lebt, wurde rekrutiert. Zur Bestimmung von BMI, taille et des hanches ont été utilisées pour déterminer IMC et le
Taille-Hüft-Quotient und Taillenumfang wurden Körpergrösse rapport taille/hanche.
und -gewicht, Taillen- und Hüftumfang erfasst. Résultats: Parmi les femmes ayant un IMC normal, 19 % avaient
Ergebnisse: Unter den Frauen mit normalem BMI hatten 19 % une circonférence de la taille élevée (≥ 80 cm) et 8,5 % avaient
einen grossen Taillenumfang (≥ 80 cm) und 8,5 % hatten einen un rapport taille/hanche élevé (> 0.85). Parmi les hommes ayant
hohen Taille-Hüft-Quotient (> 0,85). Unter den Männern mit un IMC normal, 3,4 % avaient une circonférence de la taille
normalem BMI hatten 3,4 % einen grossen Taillenumfang élevée (≥ 95 cm) et 0,1 % avaient un rapport taille/hanche élevé
(≥ 95 cm) und 0,1 % einen hohen Taille-Hüft-Quotient (> 1,0). (> 1).
Schlussfolgerung: Alle Messgrössen, BMI, Taille-Hüft-Quotient Conclusions: L’indice de masse corporelle, le rapport taille/
und Taillenumfang müssen bei der Identifikation von überge- hanche et la circonférence de la taille jouent tous un rôle dans
wichtigen und adipösen Personen berücksichtigt werden. l’identification des personnes obèses ou ayant du surpoids.

References Centre for Population Studies in Epidemiology Defay R, Delcourt C, Ranvier M, Lacroux A,
(2002a). The North West Adelaide Health Study Papoz L, POLA Study Group (2001). Relations-
Assayama K, Oguni T, Hayashi K, et al. (2000). – April 2002. Adelaide: Department of Human hips between physical activity, obesity and dia-
Critical value for the index of body fat distribu- Services. betes mellitus in a French elderly population: the
tion based on waist and hip circumferences and http://www.dhs.sa.gov.au/pehs/CPSE/brief- POLA study. Int J Obesity 25: 512–8.
stature in obese girls. Int J Obesity 24: 1026–31. reports-cpse.html (accessed April 2003)
Despres J-P, Lemieux I, Prud’homme D (2001).
Australian Bureau of Statistics (2000). Popula- Centre for Population Studies in Epidemiology Treatment of obesity: need to focus on high risk
tion by age and sex. South Australia. 30 June (2002b). The North West Adelaide Health Study: abdominally obese patients. BMJ 322: 716–20.
1999. Canberra: Australian Bureau of Statistics. brief report 2002–08. Adelaide: Department of
Human Services. Deurenberg P, Deurenberg YM, Wang J, Lin F,
Australian Centre for Diabetes Strategies (2000). http://www.dhs.sa.gov.au/pehs/CPSE/NWhealth. Schmidt G (1999). The impact of body build on
National Evidence-Based Guidelines for the html (accessed April 2003) the relationship between body mass index and
Management of Type 2 Diabetes Mellitus. Rev. percent body fat. Int J Obesity 23: 537–42.
Draft. Sydney: Diabetes Australia Guideline Cole T, Bellizzi M, Flegal K, Dietz W (2000).
Development Consortium. Establishing a standard definition for child over- Doll S, Paccaud F, Bovet P, Burnier M, Wietlis-
weight and obesity worldwide: international bach V (2002). Body mass index, abdominal adi-
Bandini L, Dietz W (1987). Assessment of body survey. BMJ 320: 1–6. posity and blood pressure: consistency of their
fatness in childhood obesity: evaluation of labo- association across developing and developed
ratory and anthropometric techniques. J Am Dean A, Dean J, Coulombier D, et al. (1994). Epi countries. Int J Obesity 26: 48–57.
Diet Assoc 87: 1344–8. Info Version 6: a word processing, database, and
statistics program for epidemiology on micro- Han T, van Leer E, Seidell J, Lean M (1995).
Booth M, Hunter C, Gore C, Bauman A, Owen N computers. Kerr AG, ed. Atlanta: Centers for Waist circumference action levels in the identifi-
(2000). The relationship between body mass Disease Control and Prevention. cation of cardiovascular risk factors: prevalence
index and waist circumference: implications for study in a random sample. BMJ 311: 1401–5.
estimates of the population prevalence of over-
weight. Int J Obesity 24: 1058–61.

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
198 Section: International comparison of health determinants Gill T, Chittleborough C, Taylor A, et al.
Body mass index, waist hip ratio, and waist circumference

Ho S, Chen Y, Woo J, Leung S, Lam T, Janus E Molarius A, Seidell J, Sans S, Tuomilehto J, Kuu- Visscher T, Seidell J, Molarius A, van der Kuip D,
(2001). Association between simple anthropome- lasmaa K, for the WHO MONICA Project Hofman A, Witteman J (2001). A comparison of
tric indices and cardiovascular risk factors. Int J (1999). Waist and hip circumferences, and waist- body mass index, waist-hip ratio and waist cir-
Obesity 25: 1689–97. hip ratio in 19 populations of the WHO cumference as predictors of all-cause mortality
MONICA Project. Int J Obes 23: 116–125. among the elderly: the Rotterdam study. Int J
Hu F, Wang B, Chen C, et al. (2000). Body mass Obesity 25: 1730–5.
index and cardiovascular risk factors in a rural National Heart Foundation (1989). Risk Factor
Chinese population. Am J Epidemiol 151: 88–97. Prevalence Study No. 3. Canberra: National Wilson D, Wakefield MA, Taylor A (1992). The
Heart Foundation. South Australian Health Omnibus Survey.
International Agency for Research on Cancer Health Promot J Aust 2:47-49.
(2001). Biennial Report 2000 – 2001. Lyon: Perissinotto E, Pissent C, Sergi G, Grigoletto F,
IARC. Enzi G, ILSA Working Group (2002). Anthropo- World Health Organization (1998). Obesity: pre-
metric measurements in the elderly: age and venting and managing the global epidemic.
Lean M, Han T, Morrison C (1995). Waist cir- gender differences. Br J Nutr 87: 177–86. Report of a WHO consultation on obesity,
cumference as a measure for indicating need for Geneva, 3–5 1997. Geneva: WHO.
weight management. BMJ 311: 158–61. Pilotto L, Smith B, Nitschke M, Ruffin R, Mitchell
R (1999). Industry, air quality, cigarette smoke
Lev-Ran A (2001). Human obesity: an evolutio- and rates of respiratory illness in Port Adelaide.
nary approach to understanding our bulging Aust NZ J Public Health 23: 657–60.
waistline. Diabetes Metab Res Rev 17: 347–62.
Address for correspondence
Reilly J, Dorosty A, Emmett P, The ALSPAC
Malina R, Katzmarzyk P (1999). Validity of the Study Team (2000). Identification of the obese
body mass index as an indicator of the risk and child: adequacy of the body mass index for cli- Tiffany Gill
presence of overweight in adolescents. Am J Clin nical practice and epidemiology. Int J Obesity 24: Population Research and
Nutr 70(suppl): 131S–6S. 1623–7. Outcome Studies Unit
Level 8, Citicentre Building
McCarthy S, Harrington K, Kiely M, et al. (2001). Roubenoff R, Dallal G, Wilson P (1995). Predic- 11 Hindmarsh Square
Analyses of the anthropometric data from the ting body fatness: the body mass index vs estima- Adelaide, South Australia, 5000
North/South Ireland Food Consumption Survey. tion by bioelectrical impedance. Am J Public Australia
Public Health Nutr 4: 1099–106. Health 85: 726–8. Tel.: +61 8 8226 0739
e-mail: tiffany.gill@dhs.sa.gov.au
Molarius A, Seidell J (1998). Selection of anthro- Statistical Package for the Social Sciences Inc
pometric indicators for classification of abdo- (1999). SPSS Advanced Statistics 10.0. Chicago:
minal fatness – A critical review. Int J Obes 22: SPSS Inc.
719–727.

Appendix

Table A1 Body mass index (BMI), waist hip ratio (WHR) and waist circumference (WC) among males, North West Adelaide, South Australia, 2000
(n = 1 236)

Age group n Percentiles (P) Mean (SD)

(years)
P10 P25 P50 P75 P90

BMI (kg/m2)
18 – 19 55 19.3 20.0 22.3 25.2 27.1 23.0 (3.1)
20 – 24 165 20.3 21.4 24.5 27.6 33.4 25.2 (4.9)
25 – 29 130 20.7 23.5 26.1 31.2 35.5 27.9 (6.4)
30 – 34 120 21.4 23.6 26.9 30.5 35.1 27.9 (5.5)
35 – 39 106 23.4 24.7 26.8 30.6 33.9 27.6 (4.3)
40 – 44 131 23.5 25.9 28.2 31.2 36.6 29.0 (5.0)
45 – 49 97 23.4 25.3 28.0 29.8 33.1 27.9 (3.5)
50 – 54 98 24.7 26.1 28.4 31.2 35.8 29.0 (4.4)
55 – 59 67 23.9 25.7 29.2 33.2 38.4 30.0 (5.8)
60 – 64 71 24.0 26.2 28.5 30.9 34.0 28.9 (4.3)
65 – 69 61 24.2 26.2 28.7 30.9 35.5 29.2 (4.9)
70 – 74 57 24.1 26.2 28.1 29.7 32.7 28.0 (4.1)
75+ 78 21.3 23.4 26.3 28.5 30.3 26.0 (3.7)

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
Gill T, Chittleborough C, Taylor A, et al. Section: International comparison of health determinants 199
Body mass index, waist hip ratio, and waist circumference

Table A1 (continued)

Age group n Percentiles (P) Mean (SD)

(years)
P10 P25 P50 P75 P90

WHR
18 – 19 55 0.80 0.81 0.84 0.89 0.89 0.84 (0.04)
20 – 24 165 0.77 0.80 0.86 0.90 0.92 0.86 (0.06)
25 – 29 130 0.81 0.84 0.88 0.92 0.99 0.89 (0.07)
30 – 34 120 0.84 0.87 0.90 0.95 0.98 0.91 (0.05)
35 – 39 106 0.86 0.88 0.92 0.94 0.97 0.92 (0.05)
40 – 44 131 0.89 0.91 0.94 0.98 1.00 0.94 (0.05)
45 – 49 97 0.87 0.89 0.93 0.96 0.99 0.93 (0.05)
50 – 54 98 0.87 0.91 0.95 0.98 1.03 0.95 (0.06)
55 – 59 67 0.89 0.92 0.95 0.99 1.03 0.96 (0.05)
60 – 64 71 0.90 0.92 0.95 0.99 1.02 0.95 (0.05)
65 – 69 61 0.89 0.92 0.95 0.99 1.02 0.95 (0.05)
70 – 74 57 0.89 0.92 0.95 0.99 1.02 0.95 (0.05)
75 + 78 0.86 0.91 0.94 0.96 1.00 0.93 (0.05)
WC
18 – 19 55 76.0 78.0 82.2 87.0 89.0 82.5 (6.2)
20 – 24 165 71.5 77.7 87.0 97.0 106.5 87.6 (13.6)
25 – 29 130 76.2 83.4 91.0 104.5 113.0 94.0 (17.0)
30 – 34 120 79.6 84.0 93.0 102.0 112.0 94.5 (12.6)
35 – 39 106 83.5 88.0 94.0 102.0 110.0 94.9 (10.2)
40 – 44 131 87.2 93.0 98.0 106.1 119.0 100.3 (12.3)
45 – 49 97 85.0 90.0 97.6 103.0 112.1 97.3 (9.7)
50 – 54 98 85.7 93.0 99.7 107.3 115.8 101.0 (12.0)
55 – 59 67 87.9 94.9 101.4 112.6 124.6 104.7 (15.2)
60 – 64 71 89.6 94.5 99.3 106.5 113.9 101.5 (10.9)
65 – 69 61 87.0 95.1 101.0 108.0 116.9 102.5 (14.0)
70 – 74 57 87.0 94.0 100.7 108.4 116.1 101.0 (10.4)
75 + 78 85.4 89.7 96.9 103.9 108.5 96.7 (10.0)

Table A2 Body mass index (BMI), waist hip ratio (WHR) and waist circumference (WC) among females, North West Adelaide, South Australia, 2000
(n = 1 287)

Age group n Percentiles (P) Mean (SD)

(years)
P10 P25 P50 P75 P90

BMI (kg/m2)
18 – 19 90 20.5 21.7 23.6 26.4 40.0 25.8 (6.5)
20 – 24 118 18.8 20.8 22.2 29.5 30.9 23.9 (4.8)
25 – 29 116 19.7 21.6 24.8 32.2 36.3 26.8 (6.3)
30 – 34 126 19.7 21.4 23.9 29.9 35.7 25.9 (6.1)
35 – 39 100 21.3 22.7 26.5 31.2 36.3 27.6 (5.8)
40 – 44 134 20.7 22.8 26.4 31.5 37.8 28.2 (7.1)
45 – 49 93 21.8 24.1 28.1 33.8 37.5 29.2 (6.7)
50 – 54 106 22.6 24.8 27.9 32.7 37.0 29.2 (6.0)
55 – 59 80 21.9 24.5 28.1 31.9 36.1 28.6 (5.6)
60 – 64 66 23.3 25.5 28.3 31.2 37.1 28.9 (4.8)
65 – 69 70 22.1 24.5 28.2 32.5 36.3 28.7 (5.7)
70– 74 63 22.2 25.2 28.5 31.4 36.9 28.9 (5.5)
75 + 122 22.2 25.0 28.2 31.1 34.1 28.1 (4.5)
WHR
18 – 19 90 0.71 0.74 0.77 0.79 0.87 0.78 (0.05)
20 – 24 118 0.69 0.72 0.76 0.79 0.83 0.76 (0.05)
25 – 29 116 0.71 0.73 0.77 0.81 0.88 0.78 (0.07)
30 – 34 126 0.73 0.74 0.78 0.82 0.86 0.79 (0.05)
35 – 39 102 0.74 0.76 0.79 0.83 0.88 0.80 (0.06)
40 – 44 134 0.73 0.75 0.80 0.86 0.89 0.81 (0.07)
45 – 49 93 0.74 0.77 0.81 0.86 0.90 0.82 (0.06)
50 – 54 106 0.75 0.78 0.82 0.86 0.90 0.82 (0.06)
55 – 59 80 0.74 0.77 0.82 0.88 0.93 0.83 (0.07)
60 – 64 66 0.75 0.78 0.82 0.89 0.93 0.83 (0.07)
65 – 69 70 0.75 0.78 0.83 0.88 0.92 0.83 (0.07)
70 – 74 63 0.77 0.81 0.85 0.88 0.94 0.85 (0.06)
75 + 122 0.77 0.81 0.85 0.88 0.92 0.85 (0.06)

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003
200 Section: International comparison of health determinants Gill T, Chittleborough C, Taylor A, et al.
Body mass index, waist hip ratio, and waist circumference

Table A2 (continued)

Age group n Percentiles (P) Mean (SD)

(years)
P10 P25 P50 P75 P90

WC
18 – 19 90 65.0 71.0 73.0 78.2 103.0 77.4 (12.5)
20 – 24 118 65.0 67.0 72.5 80.2 93.7 75.5 (10.8)
25 – 29 116 66.0 70.0 79.0 89.3 100.8 82.0 (14.3)
30 – 34 126 68.0 71.0 78.0 88.0 102.0 81.0 (13.5)
35 – 39 102 69.7 73.0 83.8 94.0 104.3 85.0 (14.0)
40 – 44 134 68.2 74.7 82.7 98.5 108.5 86.7 (15.7)
45 – 49 93 70.7 77.0 87.9 97.4 106.9 88.9 (14.9)
50 – 54 106 73.0 78.2 86.9 99.0 110.0 89.5 (14.0)
55 – 59 80 71.5 78.0 89.7 99.5 106.0 89.7 (14.1)
60 – 64 66 75.9 80.7 87.7 98.8 111.6 90.2 (12.7)
65 – 69 70 73.8 81.7 89.5 100.2 109.5 90.7 (13.2)
70 – 74 63 77.0 83.0 91.0 100.3 110.3 92.3 (12.3)
75 + 122 77.0 84.0 93.0 99.5 107.7 91.7 (11.4)

Soz.- Präventivmed. 48 ( 2003) 191 – 200

© Birkhäuser Verlag, Basel, 2003