VOLUME 19 NUMBER 1 | jaNUaRy 2011 | www.obesityjournal.org
nature publishing group
The nature of excess body weight may be changing to one of
greater central adiposity (1,2). However, it is unclear if this is
consistent across population subgroups and body weights. As
abdominal obesity appears more strongly correlated with cardio-
vascular risk than BMI (3,4) changes to greater abdominal obes-
ity have important implications for the burden of obesity-related
disease. Comparing recent trends in BMI and waist circumfer-
ence (WC) will help to elucidate whether the nature of excess
body weight is changing over time, and provide insight into
whether BMI is capturing risk as well as generally accepted.
The aim of this study is to analyze the degree to which
abdominal overweight is increasing beyond what would be
expected from changes in BMI alone. In particular, we have
studied these differential changes within population subgroups
determined by age, sex, education, race-ethnicity, and body
weight, and suggest possible public health implications of the
findings. We analyzed changes in WC and BMI between 1988–
1994 and 2005–2006 using comparable, nationally representa-
tive cross-sectional NHANES surveys of the United States.
This study examined data from two cross-sectional surveys of the US
population, the National Health and Nutrition Examination Studies
(NHANES) (3), conducted from 1988 to 1994 (NHANES III) and
2005 to 2006. Participants were selected using a stratified cluster
sampling design representative of the civilian, noninstitutionalized
US population. After being interviewed in their homes, participants
were invited for a clinical examination at a mobile examination centre,
where anthropometric data were collected. Only participants with data
from the mobile examination centre were included in the analysis. The
household interview response rates and mobile examination centre
response rates in the respective surveys were 86 and 78% in NHANES
III (1988–1994) and 80 and 77% in 2005–2006. The sample design and
weighting methodology was similar in both surveys and the studies
included 31,311 and 9,950 individuals, respectively (3,4). We excluded
participants with missing height, weight, WC, and education data,
those aged <20 years, and pregnant women, leaving 15,349 and 4,176
records for analysis.
WC was measured to the nearest 0.1 cm at the high point of the iliac crest
at minimal respiration. Weight was measured on a self-zeroing weight
scale (Toledo Scale; Toledo, Columbus, OH). Height was measured to
the nearest millimeter (5).
WC cut-points classified people into “low risk” (men <94 cm; women
<80 cm), “increased risk” (men ≥94–<102 cm; women ≥80–<88 cm),
and “substantially increased risk” (men ≥102 cm; women ≥88 cm)
(6). BMI was calculated as weight divided by height squared (kg/m2).
Standard cut-points applied to BMI scores classified people into under-
weight (<18.5 kg/ m2), normal weight (18.5–24.99 kg/m2), overweight
(25–29.99 kg/m2), and obese (≥30 kg/m2) (7).
1Department of Epidemiology and Preventive Medicine, Monash University, Alfred Hospital, Victoria, Australia. Correspondence: Helen L. Walls
Received 25 November 2009; accepted 24 May 2010; published online 17 June 2010. doi:10.1038/oby.2010.149
Comparing Trends in BMI and Waist
Helen L. Walls1, Christopher E. Stevenson1, Haider R. Mannan1, Asnawi Abdullah1, Christopher M.
Reid1, John J. McNeil1 and Anna Peeters1
The nature of excess body weight may be changing over time to one of greater central adiposity. The aim of this
study is to determine whether BMI and waist circumference (WC) are increasing proportionately among population
subgroups and the range of bodyweight, and to examine the public health implications of the findings. Our data are
from two cross-sectional surveys (the US National Health and Nutrition Examination Studies (NHANES) in 1988–1994
(NHANES III) and 2005–2006), from which we have used samples of 15,349 and 4,176 participants aged ≥20 years.
Between 1988–1994 and 2005–2006 BMI increased by an average of 1.8 kg/m2 and WC by 4.7 cm (adjusted for sex,
age, race-ethnicity, and education). The increase in WC was more than could be attributed simply to increases in BMI.
This independent increase in WC (of on average, 0.9 cm) was consistent across the different BMI categories, sexes,
education levels, and race-ethnicity groups. It occurred in younger but not older age groups. Overall in each BMI
category, the prevalence of low-risk WC decreased and the prevalence of increased-risk or substantially increased-
risk WC increased. These results suggest that the adverse health consequences associated with obesity may be
increasingly underestimated by trends in BMI alone. Since WC is closely linked to adverse cardiovascular outcomes,
it is important to know the prevailing trends in both of these parameters.
Obesity (2011) 19, 216–219. doi:10.1038/oby.2010.149
obesity | VOLUME 19 NUMBER 1 | jaNUaRy 2011 217
In addition to BMI and WC, other variables included in this analysis are
sex, age group, education, and race-ethnicity.
We calculated the mean WC and BMI in each survey. Analyses were
stratified by sex, age group, and race-ethnicity and analyzed in SAS
(SAS version 9.1.3; SAS Institute, Cary, NC). Linear regression exam-
ined changes in the level of WC for a given level of BMI. The analysis
accounted for the complex, multistage, cluster sampling design by
using the Taylor expansion method to estimate sampling errors of
Mean age and sex were similar across the surveys, as were the
response rates. Overall, the proportion of non-Hispanic whites
increased and the proportion of non-Hispanic blacks and
Mexican Americans decreased. The proportion completing
high school increased.
Between 1988–1994 and 2005–2006, increases in both BMI
and WC were observed. Mean BMI increased from 27.1 to
28.7 kg/m2, and mean WC increased from 93.3 to 98.1 cm.
After adjusting for sex, age, race-ethnicity, and education,
BMI increased between 1988–1994 and 2005–2006 by 1.8 kg/ m2
and WC by 4.7 cm. After additional adjustment for BMI, WC
increased by 0.9 cm (95% confidence interval 0.5–1.2). After
adjusting for WC, there was no significant change in BMI
(−0.0 kg/m2; 95% confidence interval −0.2 to 0.1).
There were independent increases in WC in all three BMI
categories (Table 1). On average in 2005–2006 individuals in
the normal-weight, overweight, and obese groups had WC 0.6,
0.8, and 1.0 cm larger than those in 1988–1994. There were
similar independent WC increases in both sexes, education
levels, and in the three most common race-ethnicity groups.
Interactions with age group were statistically significant. There
were independent increases in WC in those aged 20–49 years
but not those aged 50+ years. The consequences of this was
that on average in 2005–2006 individuals aged 20–29, 30–39,
and 40–49 years has waists 1.8, 1.2, and 0.7 cm larger than their
counterparts in 1988–1994.
Within each BMI category the proportion of higher-risk WC
increased for males and females aged <50 years. In 1988–1994
table 1 Changes in waist circumference independent of BMI between 1988–1994 and 2005–2006 using linear regression
Change in WC (cm)
(95% CI) over and above
changes in BMI
Total population 0.86 (0.53–1.19) 93.33 (92.97–93.69) 98.08 (97.01–99.16)
Normal-weight population (BMI 18.5–25 kg/m2) 0.62 (0.25–0.98)b
81.71 (81.30–82.11) 83.25 (82.71–83.78)
Overweight population (BMI 25–30 kg/m2) 0.75 (0.29–1.16)b
95.12 (94.78–95.46)96.63 (96.00–97.25)
Obese population (BMI ≥30 kg/m2) 0.95 (0.43–1.48)b
109.69 (109.32–110.06) 113.14 (112.19–114.10)
85.30 (84.69–85.90)91.35 (89.65–93.06)
91.30 (90.57–92.04) 96.43 (94.89–97.97)
40–49 0.72 (0.15–1.28)b
95.27 (94.66–95.89) 99.19 (97.63–100.74)
50–59−0.02 (−0.75 to 0.70) 97.79 (97.05–98.54)100.60 (98.98–102.22)
60–69 −0.25 (−0.23 to 0.73) 98.89 (98.28–99.50)102.75 (101.21–104.29)
70+0.67 (−0.12 to 1.47) 96.00 (95.52–96.48)99.37 (97.86–100.89)
95.38 (94.95–95.81)100.80 (99.48–102.12)
91.44 (90.92–91.96)95.22 (94.08–96.35)
Did not complete high school0.68 (0.18–1.17)b
95.20 (94.70–95.69)98.64 (97.65–99.62)
Completed high school0.92 (0.60–1.25)b
92.07 (91.65–92.50)97.88 (96.66–99.10)
Non-Hispanic white0.89 (0.33–1.45)b
93.69 (93.15–94.24) 98.48 (96.98–99.98)
Non-Hispanic black0.50 (0.13–0.87)b
93.11 (92.50–93.71)99.29 (98.17–100.42)
Mexican American0.95 (0.52–1.39)b
93.53 (92.80–94.26)97.18 (96.13–98.24)
Other race (including mixed race)0.93 (−0.60 to 2.46) 87.11 (84.93–89.29)92.91 (90.06–95.75)
Other Hispanic0.88 (−0.07 to 1.82)91.19 (90.13–92.26)94.86 (92.60–97.13)
Adjustment for age, sex, race-ethnicity, education, and BMI.
CI, confidence interval; WC, waist circumference.
aChanges differ significantly by age group. bSignificant increase in WC independent of BMI.
VOLUME 19 NUMBER 1 | jaNUaRy 2011 | www.obesityjournal.org
of those aged <50 years in the normal-weight BMI category,
3.0% of men and 29.5% of women had an increased-risk or
substantially increased-risk WC, but in 2005–2006 this had
increased to 5.9 and 43.0%, respectively. Similarly, whereas in
1988–1994 the proportion of nonobese (BMI < 30) males and
females aged <50 years with substantially-increased-risk WC
was 5.1 and 25.2%, by 2005–2006 this had increased to 11.5
and 33.4%, respectively. The proportion of obese males and
females aged <50 years without a substantially-increased-risk
WC decreased from 11.6 and 1.9% in 1988–1994 to 6.9 and
0.6% in 2005, respectively.
Between 1988–1994 and 2005–2006 there was an independent
average increase of 0.9 cm in the WC of American adults over
and above that for BMI. This independent increase occurred
in all BMI categories, in both sexes, education levels and in
the three most common race-ethnicity groups. However, it
occurred in younger but not older people.
The consequence of these differential changes in weight
and WC was that within each of the normal-weight, over-
weight, and obese BMI categories the proportion of those
with higher-risk WC increased for males and females aged
<50 years. The apparent greater association of WC than BMI
with metabolic risk (3,4) may mean that overall in each BMI
category, the prevalence of having a low-risk WC decreased
and the prevalence of an increased-risk or substantially
increased-risk WC increased. These changes affected the
normal-weight category as well as the overweight and obese
Our findings are consistent with a number of studies that
have suggested an independent increase in WC over and
above that for BMI (1,9–12). However, some of these studies
are based on differences in the percentage change in BMI and
WC, which does not inform whether one indicator has moved
independently further than the other, due to the nature of the
relationship between BMI and WC. In our analysis, while there
was a marginally greater percentage change in BMI (5.9%)
than WC (5.1%), we demonstrated that increases in WC were
greater than expected based on increases in BMI. Elobeid et al.
conducted the only study to specifically look at increases in
WC independent of increases in BMI (1). The strengths of our
study are that we looked at the increase in WC independent of
increases in BMI over a range of demographic groups, and that
we quantified these absolute changes in WC.
There were no differences in sampling design between the
surveys and the survey response rates were similar, reduc-
ing the impact of bias from these sources. The discrepancy
between the total sample size and the number of participants
for whom data were analyzed—due to participants missing
height, weight, WC, and/or education variables—is likely
to incur bias. Nonresponse has been associated with having
a higher body weight and lower socioeconomic status (also
associated with a higher body weight in high-income coun-
tries) (13,14). However, it is unlikely that this bias would
affect the key finding of this paper of an independent increase
of WC over and above the increase in BMI. That the mean
WC values are close to the cut-points for abdominal obesity
could explain why a small increase in WC could lead to a large
increase in the prevalence of increased risk WC (9). However,
this appears not to be the case in this study as we have used
linear regression to explore the continuous changes in WC
over and above BMI.
Changing ethnic makeup may explain the changes in
body fat distribution; however, we obtained similar results
after stratifying by race-ethnicity. Other possible con-
tributors include declining physical activity levels, sleep
deprivation, stress, endocrine disruptors, and certain
The changes occurred for all races, education levels, sexes,
and body weights, including in people considered “normal
weight.” The nature of the changes suggest that using solely
BMI to monitor the impact of increasing weight status in
the population may lead to an underestimation of the asso-
ciated health burden. That the independent increase in WC
has been shown in people aged <50 years but not older age
groups suggests that the changes in our environment have
impacted more on younger adults, or that they are the result
of more recent environmental changes.
The findings suggest that within each BMI category the
mean WC has increased. To date these independent changes
have been restricted to younger people. Considering the asso-
ciation between WC and the level of metabolic risk, the sole
use of BMI to monitor the impact of increasing population
weight may lead to an underestimation of the obesity-related
We thank the NHaNES investigators for the use of their dataset. Financial
support was received from the National Health and Medical Research
Council (scholarship H.L.W.; grant no. 465130 H.L.W., C.S., H.R.M.),
National Heart Foundation (H.L.W.) ausaID (a.a.), and VicHealth (a.P.).
The authors declared no conflict of interest.
© 2010 The Obesity Society
1. Elobeid MA, Desmond RA, Thomas O, Keith SW, Allison DB. Waist
circumference values are increasing beyond those expected from BMI
increases. Obesity (Silver Spring) 2007;15:2380–2383.
2. Walls HL, Wolfe R, De Courten M et al. Waist Circumference Trends in
Urban Australian Adults, 1989–2000. Australia New Zealand Society of
Obesity: Canberra, 2007.
3. Dalton M, Cameron AJ, Zimmet PZ et al.; AusDiab Steering Committee.
Waist circumference, waist-hip ratio and body mass index and their
correlation with cardiovascular disease risk factors in Australian adults.
J Intern Med 2003;254:555–563.
4. Zhu S, Heshka S, Wang Z et al. Combination of BMI and Waist
Circumference for Identifying Cardiovascular Risk Factors in Whites.
Obes Res 2004;12:633–645.
5. Li C, Ford ES, Mokdad AH, Cook S. Recent trends in waist circumference
and waist-height ratio among US children and adolescents. Pediatrics
6. Lean ME, Han TS, Morrison CE. Waist circumference as a measure for
indicating need for weight management. BMJ 1995;311:158–161.
7. WHO. Obesity: Preventing and Managing the Global Epidemic. World Health
Organization: Geneva, 2000.
obesity | VOLUME 19 NUMBER 1 | jaNUaRy 2011 219 Download full-text
8. Fuller W. Regression analysis for sample survey. Sankhya: The Indian
J Stat C 1975;37:117–132.
9. Li C, Ford ES, McGuire LC, Mokdad AH. Increasing trends in waist
circumference and abdominal obesity among US adults. Obesity
(Silver Spring) 2007;15:216–224.
10. Lahti-Koski M, Pietinen P, Männistö S, Vartiainen E. Trends in waist-to-hip
ratio and its determinants in adults in Finland from 1987 to 1997. Am J Clin
11. Lahti-Koski M, Harald K, Männistö S, Laatikainen T, Jousilahti P. Fifteen-year
changes in body mass index and waist circumference in Finnish adults.
Eur J Cardiovasc Prev Rehab 2007;14:398–404.
12. Wardle J, Boniface D. Changes in the distributions of body mass index and
waist circumference in English adults, 1993/1994 to 2002/2003. Int J Obes
13. Van Loon AJ, Tijhuis M, Picavet HS, Surtees PG, Ormel J. Survey
non-response in the Netherlands: effects on prevalence estimates and
associations. Ann Epidemiol 2003;13:105–110.
14. Hill A, Roberts J, Ewings P, Gunnell D. Non-response bias in a lifestyle
survey. J Public Health Med 1997;19:203–207.
15. Keith SW, Redden DT, Katzmarzyk PT et al. Putative contributors to the
secular increase in obesity: exploring the roads less traveled. Int J Obes