Prevalence of class-I, class-II and class-III obesity in Australian adults between 1995 and 2011-12

O

O

O

Po1

CEBA

a

Ab

c

Ad

e

Uf

Z

R

T

h1

ARTICLE IN PRESSRCP-421; No. of Pages 10

besity Research & Clinical Practice (2015) xxx, xxx—xxx

RIGINAL ARTICLE

revalence of class-I, class-II and class-IIIbesity in Australian adults between995 and 2011—12

atherine Keatinga,b,∗, Kathryn Backholera,mma Gearona,c, Christopher Stevensond,oyd Swinburne,f, Marj Moodieb, Rob Carterb,nna Peetersa

Obesity and Population Health, Baker IDI Heart and Diabetes Institute, Melbourne,ustraliaDeakin Health Economics, Deakin University, Melbourne, AustraliaSchool of Public Health and Preventative Medicine, Monash University, Melbourne,ustraliaSchool of Health and Social Development, Deakin University, Melbourne, AustraliaWorld Health Organisation Collaborating Centre for Obesity Prevention, Deakinniversity, Melbourne, AustraliaPopulation Nutrition and Global Health, University of Auckland, Auckland, Newealand

eceived 12 September 2014; received in revised form 21 November 2014; accepted 8 February 2015

KEYWORDSObesity;Prevalence;Trends;

SummaryObjective: To compare the prevalence of class-I, II and III obesity in Australian adultsbetween 1995, 2007—08 and 2011—12.Methods: Prevalence data for adults (aged 18+ years) were sourced from cus-

Please cite this article in press as: Keating C, et al. Prevalence of class-I, class-II and class-III obesity in Australianadults between 1995 and 2011—12. Obes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.orcp.2015.02.004

Australia;Socioeconomic

tomised data from the nationally representative National Nutrition Survey (1995),the National Health Survey (2007—08), and the Australian Health Survey (2011—12)conducted by the Australian Bureau of Statistics. Obesity classifications were basedon measured height and weight (class-I body mass index: 30.0—34.9 kg/m2, class-II:35.0—39.9 kg/m2 and class-III: ≥40.0 kg/m2). Severe obesity was defined as class-IIor class-III obesity.

∗ Corresponding author at: Baker IDI Heart and Diabetes Institute, Level 4, 99 Commercial Road, Melbourne, VIC 3004, Australia.el.: +61 03 8532 1866; fax: +61 03 8532 1100.

E-mail addresses: [email protected], [email protected] (C. Keating).

ttp://dx.doi.org/10.1016/j.orcp.2015.02.004871-403X/© 2015 Asian Oceanian Association for the Study of Obesity. Published by Elsevier Ltd. All rights reserved.

dx.doi.org/10.1016/j.orcp.2015.02.004

dx.doi.org/10.1016/j.orcp.2015.02.004

mailto:[email protected]

mailto:[email protected]

dx.doi.org/10.1016/j.orcp.2015.02.004

ARTICLE INORCP-421; No. of Pages 10

2

Results: Between 1995 and 2increased from 19.1% to 27.2II and III obesity were 1.3, 1.7class I, II and III obesity was 16.9 and 4.2 per cent respecobese, increasing from one

represented in this populatioof area-level socioeconomic

classes. Severe obesity affecquintiles respectively.Conclusion: Over the last twprevalence of obesity, particlights high risk groups who w© 2015 Asian Oceanian Asso

ihasttr

gdhii[nps

robswei

M

D

Ltd. All rights reserved.

Introduction

The World Health Organisation defines three sub-classes of obesity severity (class-I body mass index(BMI): 30.0—34.9 kg/m2, class-II: 35.0—39.9 kg/m2

and class-III: ≥40.0 kg/m2) [1]. Severe obesity isdefined as class-II or class-III obesity. We havepreviously reported that between 1980 and 2000,the prevalence of obesity (all classes) increasedfrom 10.6 to 17.2% in men and 8.8 to 19.3% inwomen aged 25—65 in Australia [2]. Increases weregreatest in the more severe obesity sub-classes.During this period, the prevalence of class-III obe-sity increased 4.8-fold in women and 2.5-fold inmen, whilst comparable increases in class-I obesitywere 1.8-fold and 1.5-fold respectively [2]. Simi-lar observations were reported in other developedcountries during the same period [3—5]. During thelast decade, the prevalence of obesity has contin-ued to increase. The Australian Bureau of Statisticsreport that 27.3% of Australians were obese (allclasses) in 2011—12, an increase from 24.0% in2007—08 [6]. To the best of our knowledge, recentobesity trends by severity sub-class have not beenexamined in Australia [2].

An exponential increase in the risk of adversehealth outcomes is observed with increasing sever-ity of obesity. For example, it has been estimatedthat the risk of developing type 2 diabetes isincreased 93-fold in women and 42-fold in men who

Please cite this article in press as: Keating C, et al. Prevaleadults between 1995 and 2011—12. Obes Res Clin Pract (201

are severely obese, relative to healthy weight coun-terparts [7,8]. Similarly, a body mass index greaterthan 40 is associated with between 6.5 and 13.7years of life lost [9]. Consequently, small increases

Pap

PRESSC. Keating et al.

011—12, the prevalence of obesity (all classes combined)%. During this 17 year period, relative increases in class I,

and 2.2-fold respectively. In 2011—12, the prevalence of9.4, 5.9 and 2.0 per cent respectively in men, and 16.1,

tively in women. One in every ten people was severelyin twenty in 1995, and women were disproportionallyn. Obesity prevalence increased with increasing levelsdisadvantage, particularly for the more severely obeseted 6.2% and 13.4% in the least and most disadvantaged

o decades, there have been substantial increases in theularly the more severe levels of obesity. This study high-arrant targeted weight gain prevention interventions.ciation for the Study of Obesity. Published by Elsevier

n the prevalence of severe obesity will probablyave a similar impact on adverse health outcomess large increases in the prevalence of class-I obe-ity. Therefore, understanding the composition ofhe obese population is critical to determininghe associated morbidity and mortality burden ofecent trends.

A socioeconomic gradient in obesity, wherereater prevalence of obesity is observed in moreisadvantaged groups, has been reported in mostigh income countries [10]. Two previous stud-es have reported a greater risk of severe obesityn more socioeconomically disadvantaged groups11,6]. However to the best of our knowledge,o previous study has explored differences in therevalence of obesity classes I, II and III [1], acrossocioeconomic strata.

In the current study, we analysed nationallyepresentative data to compare the prevalencef class-I, II and III obesity in Australian adultsetween 1995, 2007—08 and 2011—12. Obesity clas-ifications were based on measured height andeight. We also examined the age, sex and socio-conomic profile for each obesity severity sub-classn 2011—12.

ethods

ata sources

nce of class-I, class-II and class-III obesity in Australian5), http://dx.doi.org/10.1016/j.orcp.2015.02.004

revalence data for the population aged 18 yearsnd over were sourced from customised data,rovided for the purposes of this research, from

dx.doi.org/10.1016/j.orcp.2015.02.004

INORCP-421; No. of Pages 10

C

tHSBtp

aSithpyA

wSdsaim

vspAsar1A

wim2tehgo(

ae‘s(mcDlBm

tsEcr

A

Ttai1sigmdutPotPp2s

pyypgoIizcltaippw

R

Ai

ARTICLElass I—III obesity prevalence

he National Nutrition Survey (1995), the Nationalealth Survey (2007—08), and the Australian Healthurvey (2011—12) each conducted by the Australianureau of Statistics (ABS). Each survey was selectedo provide a representative sample of the Australianopulation.

In 2011—12, a total of 30,721 households werepproached to participate in the Australian Healthurvey, of which 25,080 households participatedn the survey. The sample included 31,837 par-icipants, of whom 83.5% (26,577) participated ineight and weight measurement [6]. The sam-le was 24,910 after people aged less than 18ears were excluded (personal communication,BS, 26/05/14).

In 2007—08, a total of 19,979 private dwellingsere selected in the sample for the National Healthurvey, reducing to an active sample of 17,426wellings after sample loss in the field stage. Theample included 15,779 participants aged 18 yearsnd over [12], of whom 71.4% (11,266) participatedn height and weight measurements (personal com-unication, ABS, 26/05/14).In 1995, a total of 57,633 people were inter-

iewed by the National Health Survey. Of thisample, 22,562 people were selected to alsoarticipate in the National Nutrition Survey.pproximately 13,800 (61%) people completed theurvey, of whom 98.2% participated in heightnd weight measurements [13]. The sample waseduced to 10,281 after people aged less than8 years were excluded (personal communication,BS, 26/05/14).

The process for collecting physical measuresas similar for all surveys. Interviewers used dig-

tal scales to measure weight (maximum weighteasured by scales was 150 kg in 2011—12 and

007—08 and 140 kg in 1995) and a stadiometero measure height (maximum 210 cm). Interview-rs encouraged respondents to remove shoes andeavy clothing, prior to measurement. Weight (kilo-rams) and height (centimetres) were recorded tone decimal point. BMI was calculated as weightkg) divided by height (m)2.

In the 2011—12 survey, highest educationalttainment was reported by participants and cat-gorised into three mutually exclusive categories:did not complete secondary school’, ‘completedecondary school’, ‘completed degree/diploma’reflecting most to least disadvantaged socioecono-ic position (SEP)). Survey participants were also

ategorised into Index of Relative Socio-Economic


isadvantage (IRSD) quintiles. IRSD is an area-evel SEP indicator developed by the Australianureau of Statistics. This index represents a sum-ary measure from a group of 20 variables (related

I22

PRESS3

o education, income, employment, family compo-ition, housing benefits, car ownership, ethnicity,nglish language proficiency and residential over-rowding). A lower quintile indicates an area withelatively greater disadvantage.

nalysis

he Australian Bureau of Statistics provided cus-omised data on the prevalence of obesity andssociated uncertainty statistics (95% confidencentervals and standard errors) for individuals aged8+. Survey participants excluded from this data-et included females who indicated during thenterview that they were pregnant and a very smallroup of people whose weight exceeded the maxi-um weight the scales would measure (140/150 kgepending on survey year). Sampling weights weretilised to adjust results from the sample surveyo infer results for the in-scope total population.revalence data for class-I, class-II and class-IIIbesity were provided, stratified by gender anden-year age groups for each of the three surveys.revalence rates age-standardised to the 2011—12opulation were also provided. For the most recent011—12 survey, obesity prevalence data were alsotratified by educational attainment and IRSD.

Age-standardised prevalence rates were com-ared over the long-term (1995 vs 2011—12, 17ears) and short-term (2007—08 vs 2011—12, 5ears). Age-standardised prevalence was also com-ared between the most and least disadvantagedroups for education level (‘did not complete sec-ndary school’ vs ‘completed degree/diploma’) andRSD (quintile 1 vs quintile 5). The statistical signif-cance of these comparisons was assessed using a

score and associated p value. The z score wasalculated as the difference between the preva-ence estimates divided by the standard error ofhis difference, with this standard error calculateds the square root of the sum of the squares of thendividual prevalence standard errors. Associated

values were derived using the standard normalrobability distribution. A p value less than 0.05as interpreted as statistically significant.

esults

ge and sex profile of the obese populationn 2011/12


n 2011—12, the total prevalence of obesity was7.2% (27.3% for males and 27.2% for females). In011—12, out of every 100 men, 19.4 were class-I

dx.doi.org/10.1016/j.orcp.2015.02.004

Please cite

this article

in press

as: Keating

C, et

al. Prevalence

of class-I,

class-II and

class-III obesity

in Australian

adults betw

een 1995

and 2011—

12. O

bes Res

Clin Pract

(2015), http://dx.doi.org/10.1016/j.orcp.2015.02.004

AR

TIC

LE

IN P

RE

SS

ORCP-421;

N

o. of

Pages 10

4

C. Keating

et al.

Table 1 Prevalence of class-I, class-II and class-III obesity in Australian adults, 1995, 2007—08 and 2011—12.

Prevalence* Change p value

1995 2007—08 2011—12 2011—12 vs 2007—08(5 years)

2011—12 vs 1995(17 years)

2011—12 vs2007—08

2011—12 vs1995

% 95% CI % 95% CI % 95% CI Relativechange**

Absolutechange

Relativechange**

Absolutechange

MalesObese class-I 15.4 (14.5, 16.3) 18.1 (16.7, 19.5) 19.4 (18.5, 20.3) 1.07 1.3 1.26 4.0 0.129 <0.001Obese class-II 2.9 (2.4, 3.4) 5.4 (4.5, 6.3) 5.9 (5.4, 6.4) 1.09 0.5 2.03 3.0 0.336 <0.001Obese class-III 0.7 (0.4, 1.0) 1.7 (1.3, 2.1) 2.0 (1.6, 2.4) 1.18 0.3 2.86 1.3 0.306 <0.001

FemalesObese class-I 12.7 (11.9, 13.5) 14.7 (13.4, 16) 16.1 (15.3, 16.9) 1.10 1.4 1.27 3.4 0.063 <0.001Obese class-II 4.4 (3.8, 5.0) 6.0 (5.1, 6.9) 6.9 (6.2, 7.6) 1.15 0.9 1.57 2.5 0.116 <0.001Obese class-III 2.1 (1.7, 2.5) 2.7 (2.1, 3.3) 4.2 (3.6, 4.8) 1.56 1.5 2.00 2.1 0.001 <0.001

PeopleObese class-I 14.0 (13.5, 14.5) 16.4 (15.4, 17.4) 17.7 (17.1, 18.3) 1.08 1.3 1.26 3.7 0.044 <0.001Obese class-II 3.7 (3.3, 4.1) 5.8 (5.2, 6.4) 6.4 (6.0, 6.8) 1.10 0.6 1.73 2.7 0.091 <0.001Obese class-III 1.4 (1.2, 1.6) 2.2 (1.8, 2.6) 3.1 (2.8, 3.4) 1.41 0.9 2.21 1.7 <0.001 <0.001

Source: Australian National Health Surveys (2011—12 and 2007—08) and the National Nutrition Survey (1995).* Age-standardised to the 2011/12 population.

** A 1.07 relative change is equivalent to a 1.07-fold or 7% increase in prevalence.

dx.doi.org/10.1016/j.orcp.2015.02.004

ARTICLE IN PRESSORCP-421; No. of Pages 10

Class I—III obesity prevalence 5

F esity

oIwwaco

ipdo135

oI5itry

Sp

Iiinpd

1gw4at2itlcw(

laett

C

Bsfp1a

igure 1 Prevalence of class-I, class-II/II (combined) ob

bese, 5.9 were class-II obese and 2.0 were classII-obese (Table 1). Equivalent figures for womenere 16.1, 6.9 and 4.2. Relative to men, womenere 1.2- and 2.1-fold more likely to be class-IInd class-III obese respectively. For men and womenombined, one in every four people was obese, andne in every ten people was severely obese (Fig. 1).

In women, the prevalence of class-I obesityncreased with age (starting at 18—24 years) andeaked in the 55—64 year age group (21.1%) beforeecreasing with older age. The prevalence of severebesity was approximately 8% in the age groups8—24/25—34 years, 11—12% in the age groups5—44/45—54 years and 14—15% in the age groups5—64/65—74 years.

In men, the same pattern for class-I obesity wasbserved, with the highest prevalence of class-

obesity (27.8%) also observed in the age group5—64 years. The prevalence of severe obesityncreased from 4.7% in the age group 18—24 yearso 9.7% in the age group 35—44 years and thenemained roughly steady until the age group 65—74ears (Fig. 1).

ocioeconomic position of the obeseopulation in 2011/12

n 2011—12, the prevalence of obesity (all sever-ty sub-classes) in Australian adults decreased with


ncreasing level of educational attainment (‘didot complete secondary school’: 32.5%, ‘com-leted secondary school’: 22.7% and ‘completedegree/diploma’: 19.3%) (Fig. 2).

wbii

in Australian adults by age group and sex, 2011—12.

Class-I obesity prevalence estimates were 19.7%,5.7% and 13.3% in the least to most educated sub-roups. Corresponding figures for class-II obesityere 8.6%, 4.8% and 4.3%; and for class-III obesity.2, 2.3 and 1.7%. The relative risks of class-I, IInd III obesity in the least educated group, rela-ive to the most educated group were 1.5, 2.0 and.5 respectively (all p < 0.05 for prevalence compar-son between least and most educated). For men,he relative risks of class-I, II and III obesity in theeast educated group, relative to the most edu-ated group were 1.5, 2.3 and 3.3 respectively. Foromen, corresponding figures were 1.5, 1.8 and 2.4

all p < 0.05) (Fig. 2).Similar trends were observed according to area-

evel SEP (IRSD). A SEP gradient was observed forll obesity class and gender sub-groups, with thexception of class-I obese men, where a non-linearrend was observed with prevalence being rela-ively stable across the middle SEP groups (Fig. 3).

hanges in prevalence over time

etween 1995 and 2011—12, the prevalence of obe-ity (all severity sub-classes combined) increasedrom 19.1% to 27.2%. In 1995, the age-standardisedrevalence of class-I, II and III obesity were5.4, 2.9 and 0.7 per cent respectively in mennd 12.7, 4.4 and 2.1 per cent respectively in


omen (Table 1). The changes in obesity prevalenceetween 1995 and 2011—12 represented relativencreases of 1.3-, 2.0- and 2.9-fold respectivelyn men and 1.3-, 1.6- and 2.0-fold respectively in

dx.doi.org/10.1016/j.orcp.2015.02.004

ARTICLE IN PRESSORCP-421; No. of Pages 10

6 C. Keating et al.

tione su

arohtcg2owtpay21t

Figure 2 Prevalence of class I, II and III obesity by educaof this figure providing data points is provided as an onlin

women. Increases in age-standardised prevalencewere statistically significant for each obesity sub-class in both men and women (all p < 0.001).

In 2007—08, the age-standardised prevalence ofclass-I, II and III obesity were 18.1, 5.4 and 1.7per cent respectively in men and 14.7, 6.0 and 2.7per cent respectively in women (Table 1). Over thisshort time horizon, the increase in class-III obe-sity in women alone was statistically significant(p < 0.001). In all obesity class and gender sub-groupings, the 2007/8 prevalence was intermediateto the 1995 and 2012 prevalence.

Discussion


Key findings

In the current study, we compared the preva-lence of class-I, II and III obesity in Australian

soa

Figure 3 Prevalence of class I, II and III obesity by area-le*p < 0.05. A tabular version of this figure providing data point

level in Australia, 2011—12. *p < 0.05. A tabular versionpplementary table.

dults between 1995 and 2011—12. Nationally rep-esentative population samples were assessed andbesity classifications were based on measuredeight and weight. We observed that over the pastwo decades, the prevalence of all obesity sub-lasses has increased, with the greatest relativerowth in the more severe obesity sub-classes. In011—12, one in every four adults was obese andne in every ten adults was severely obese. Womenere more likely to be severely obese, including

wice as likely to be class-III obese. Severe obesityrevalence as high as 15.5% was observed in womenged 64—74 years and 10.1% in men aged 55—64ears. On a population-level, we estimate that in011—12, 2.6 million adults were class-I obese and.4 million adults were severely obese and poten-ially required treatment.


We also found that class-I, II and III obe-ity prevalence increased with increasing levelsf disadvantage based on both education andn area-level marker of socioeconomic position.

vel socioeconomic status (IRSD) in Australia, 2011—12.s is provided as an online supplementary table.

dx.doi.org/10.1016/j.orcp.2015.02.004


C

Tsaaoqsrmao

S

Ttatfuwahtwatspp

t2fautanmIbdwOasn

SbwItii

cip

ptsesttArmrfi

S

Hpins(raoalipbcpWiratTult

tcioCoo


he gradient was most pronounced in the moreeverely obese sub-classes, such that severe obesityffected 6 out of every 100 people in the least dis-dvantaged (IRSD) quintile of the population and 13ut of every 100 people in the most disadvantageduintile of the population. The prevalence of obe-ity was higher in the most disadvantaged group,elative to the least disadvantaged group, for bothen and women in all obesity severity sub-classes

ccording to education level, and all except class-Ibese men according to area-level SEP.

trengths and limitations

o the best of our knowledge, this is the first studyo examine obesity trends by severity sub-class for

large nationally representative sample popula-ion in Australia. In contrast, our previous studyor the period 1980—2000 [14] was limited to therban population aged 25—65. Height and weightere measured, survey methods were similar acrossll time points and response rates were generallyigh. Prevalence measures were age-standardisedo ensure changes in population adiposity over timeere not confounded by changes to the populationge structure. This is also the first Australian studyo describe prevalence for each obesity severityub-class stratified by age, sex and socioeconomicosition (previously not feasible due to small sam-le sizes).

The three Health Surveys measured weight upo a maximum of 140 kg in 1995 and 150 kg in011—12 and 2007—08. Weight was not recordedor a small number of participants with very highpparent weight exceeding these maximum meas-res, and these participants were excluded fromhe health survey weight data. It is possible thats the population became heavier over time, theumber of participants exceeding the maximumeasure increased, hence the prevalence of class-

II obesity at later time points reported herein maye conservative. However this effect will have beeniluted due to increases in the maximum weighthich could be measured by the scales over time.verall, the Australian Bureau of Statistics havedvised that the prevalence estimates were noteriously affected by this issue (personal commu-ication, ABS, 26/05/14).

Response rates for the 1995 National Nutritionurvey were relatively low. Potential non-responseias was reduced via the adjustment of sampleeights to reflect nationally representative data.


t is possible that obese people were less willingo participate in surveys at all time points, lead-ng to an underestimation in prevalence rates. Its also possible that lower response rates in 1995

tteo

PRESS7

ould have led to an over-estimation in the increasen obesity prevalence between 1995 and later timeoints.

With regards to the analysis of severe obesityrevalence according to socioeconomic position, tohe best of our knowledge, one previous Australiantudy has published in this theme area. Howardt al. reported that the odds ratio of severe obe-ity (class II/III combined) was 2.3 fold higher inhe most disadvantaged quintile (IRSD), relativeo the two least disadvantaged quintiles in Southustralia [11]. We observed similar results with theelative risk of class II and class III obesity in theost vs least disadvantaged quintile as 1.9 and 2.8

espectively. Therefore our results support previousndings for a nationally representative sample.

ignificance

istorically, severe obesity affected a small pro-ortion of the population in Australia (e.g. 1—2%n 1980) and has received little attention. Weow demonstrate that severe obesity affects aubstantial proportion of the adult population10%). Similarly large prevalence (15%) has beeneported for US adult populations in 2010 [15]. Thedverse health and quality of life consequencesf severe obesity for the individual are widelyccepted [16,17]. However with increasing preva-ence, adverse consequences are also becomingncreasingly large at the societal level. We havereviously demonstrated that the healthcare costsorne by the Australia government through Medi-are are greater than double in the severely obeseopulation relative to the general population [18].e have also demonstrated that the risk of disabil-

ty in the highest BMI quintile is 2—4-fold greaterelative to the lowest quintile [19], suggesting

strong relationship between increasing BMI andhe need for disability-related services and care.herefore, increases in obesity prevalence (partic-larly severe obesity) reported in this study, areikely to translate to large economic costs at bothhe individual and societal level.

The fastest growth in obesity prevalence overime was observed in the more severely obeselasses. One possible explanation for this trends that there is variation in vulnerability to thebesogenic environment across the population.onsequently, as our environment becomes morebesogenic [20], the Australian population at riskf excess weight gain is likely to gain more weight


oday than its counterparts twenty years ago. A fur-her explanation may be that for individuals withxcess weight early in life, several vicious cyclesccur whereby obesity begets obesity, leading

dx.doi.org/10.1016/j.orcp.2015.02.004


tustra[irbAtIbsi

laiHaoftsmlp

U[ptliticspooto

oaitha

ARTICLE8

to disproportionally larger weight gain over time[21]. Importantly, our analysis of multiple cross-sectional surveys does not enable conclusive resultsregarding cohort changes over time. Finally, whilethe observation of greater relative increases forsevere obesity may also in part be due to the factthat it is easier to have a greater relative increasefrom a lower baseline prevalence, prior researchhas also demonstrated greater absolute differencesin BMI over time with increasing level of BMI [14].Research to better understand the drivers of obesityprevalence patterns is urgently required to informthe most appropriate public health response.

The age distribution of severe obesity was sim-ilar for males and females and generally the mostsubstantial increases were observed between agegroups 18—24 years and 35—44 years. This sug-gests that targeting interventions in settings withyoung adults will be important for prevention. Forwomen, the prevalence of severe obesity had twonotable increases: between the age groups 25—34and 35—44 years, as well as between age groups45—54 and 55—64 years. These increases likely coin-cide with key life phases of child-bearing and theonset of menopause, highlighting these as key riskperiods for weight gain in women, and hence idealwindows for targeted intervention. Importantly thedata assessed is cross-sectional rather than cohortdata, so we cannot provide conclusive statementsabout critical periods of life. These results high-light the need to understand the drivers of the rapidincreases in severe obesity better so that we canconfirm key risk periods and sub-groups.

Policy responses to the obesity epidemic willhave a direct impact on socioeconomic disparitiesin obesity prevalence reported herein. A recentreview reported that the effectiveness of obesityprevention programs can differ by SEP, with someprograms (those focussed on education delivery)realising benefits primarily for high SEP popula-tions [13]. It is foreseeable that if delivered inisolation, these types of interventions may increasethe already large socioeconomic differences in obe-sity. It is important that we continue to implementuniversal interventions to reduce population-wideweight gain. However, the delivery of comple-mentary interventions that are tailored for, andtargeted towards high risk low SEP groups mayalso be needed to reduce socioeconomic inequal-ities in obesity and future inequalities in health.Evaluation methodologies must evolve to informdecision-making based on equity in addition to


effectiveness and cost-effectiveness.The current study demonstrated the greatest

relative socioeconomic inequalities in the moreseverely obese groups. Therefore, policy responses

nsoc


o reduce inequalities in severe obesity are partic-larly important. For this group, currently bariatricurgery is the primary available option which fos-ers large sustained weight loss [22]. We haveecently quantified socioeconomic inequalities inccess to bariatric surgery treatment in Australia23], which risks further widening the inequalitiesn severe obesity reported herein. With increasingates of severe obesity, proven treatments such asariatric surgery [22], must be more accessible toustralians from all socioeconomic groups if we areo reduce the future associated morbidity burden.ncreasing funding to treat severe obesity usingariatric surgery in public hospitals is one tangibletep which could be taken to address socioeconomicnequalities in severe obesity.

We observed significant increases in the preva-ence of class I, II and III obesity between 1995nd 2011—12, with the greatest relative increasesn class-III obesity (which more than doubled).owever, during the more recent time periodssessed (2011—12 vs 2007—08), the increase inbesity prevalence reached statistical significanceor class-III obese women only. However, the short-ime period assessed (5 years), combined with smallamples for the more severely obese categorieseans that it is not possible to identify whether

ack of statistical significance is due to insufficientower, or a plateauing in prevalence.

In the most recent decade, studies from theSA (2003—2012) [24] and the UK (1992—2010)

25] have reported a plateauing in the overallrevalence of obesity. However, it appears thathe results for overall prevalence mask ongoingarge increases at the extreme levels of obesityn both settings [15,25]. This highlights the impor-ance of stratifying by obesity severity sub-classesn epidemiological studies. It also highlights thatontinued growth in the most severe end of the obe-ity spectrum in adults appears to be a consistentattern across many countries, including Australiaver the last decade. Understanding the drivers forngoing increases in severe obesity should be priori-ised in future research to assist with the targetingf interventions to stem current trends.

Over the past two decades, the prevalence ofbesity has continued to increase in Australiandults, with the greatest relative growth observedn the most severe obesity sub-classes. One in everyen Australians is now severely obese, and evenigher rates are observed in women, older adultsnd people from the most disadvantaged socioeco-


omic groups. Continued reporting of obesity as aingle group masks the rapid increases in severebesity, a condition associated with greater adverseonsequences relative to mild obesity. Universal

dx.doi.org/10.1016/j.orcp.2015.02.004


C

essw

D

Cg1r

F

TgvsKlAacw(bfS

A

Tptti

A

Sch

R

[

[

[

[

[

[

[

[

[

[

[


fforts are required to prevent and treat mild andevere obesity in all age, sex and socioeconomicub-groups. This study highlights high risk groupsho also warrant targeted interventions.

isclosures

K previously received an independent researchrant from Allergan Australia (funding years 2010-1). EG, KB, CS, BS, MM, RC and AP declare have noelevant disclosures.

unding support

his research was supported by an ARC Linkagerant (LP120100418) and an Australian National Pre-entive Health Agency grant (188PEE2011). EG wasupported by an Australian Postgraduate Award.B was supported by a Post-doctoral Research Fel-

owship from the National Heart Foundation ofustralia (PH 12M 6824). AP was supported by

National Health and Medical Research Coun-il Career Development Fellowship (1045456). CSas supported by an ARC Discovery Project grant

DP120103277). MM, BS, RC and AP were supportedy a National Health and Research Council Centreor Research Excellence in Obesity Policy and Foodystems (#1041020).

cknowledgements

hank-you to the Australian Bureau of Statistics forroviding custom data analysed and for reviewinghe methods section of the manuscript. Thank-youo Hasini Senadheera (work experience student vis-ting Baker IDI) for reviewing the manuscript.

ppendix A. Supplementary data

upplementary data associated with this arti-le can be found, in the online version, atttp://dx.doi.org/10.1016/j.orcp.2015.02.004.

eferences

[1] World Health Organisation, 2000. Obesity: preventing andmanaging the global epidemic (WHO Technical Report Series


894) Geneva.[2] Walls HL, Wolfe R, Haby MM, Magliano DJ, de Courten M,

Reid CM, et al. Trends in BMI of urban Australian adults,1980—2000. Public Health Nutr 2010;13(5):631—8.

[

PRESS9

[3] Katzmarzyk PT, Mason C. Prevalence of class I, II and IIIobesity in Canada. CMAJ 2006;174(2):156—7.

[4] Neovius M, Teixeira-Pinto A, Rasmussen F. Shift in the com-position of obesity in young adult men in Sweden over athird of a century. Int J Obes (Lond) 2008;32(5):832—6.

[5] Sturm R. Increases in morbid obesity in the USA:2000—2005. Public Health 2007;121(7):492—6.

[6] Ljungvall A, Zimmerman FJ. Bigger bodies: long-term trends and disparities in obesity and body-massindex among U.S. adults, 1960—2008. Soc Sci Med2012;75(1):109—19.

[7] Colditz GA, Willett WC, Rotnitzky A, Manson JE. Weight gainas a risk factor for clinical diabetes mellitus in women. AnnIntern Med 1995;122(7):481—6.

[8] Chan JM, Rimm EB, Colditz GA, Stampfer MJ, Willett WC.Obesity, fat distribution, and weight gain as risk factors forclinical diabetes in men. Diabetes Care 1994;17(9):961—9.

[9] Kitahara CM, Flint AJ, Berrington de Gonzalez A, Bern-stein L, Brotzman M, MacInnis RJ, et al. Associationbetween Class III obesity (BMI of 40—59 kg/m2) and mortal-ity: a pooled analysis of 20 prospective studies. PLoS Med2014;11(7):e1001673.

10] Devaux M, Sassi F. Social inequalities in obesity andoverweight in 11 OECD countries. Eur J Public Health2013;23(3):464—9.

11] Howard NJ, Taylor AW, Gill TK, Chittleborough CR. Severeobesity: Investigating the socio-demographics within theextremes of body mass index. Obesity Res Clin Practice2008;2(1):I—II.

12] Australian Bureau of Statistics, 2013, Australian Health Sur-vey: Updated Results, 2011—2012, cat. no. 4364.0.55.003,viewed 13 August 2014, http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4364.0.55.003Explanatory%20Notes12011-2012?OpenDocument

13] Beauchamp A, Backholer K, Magliano D, Peeters A. Theeffect of obesity prevention interventions according tosocioeconomic position: a systematic review. Obesity Rev2014;15(7):541—54, http://dx.doi.org/10.1111/obr.12161.

14] Walls HL, Wolfe R, Haby MM, Magliano DJ, de Courten M,Reid CM, et al. Trends in BMI of urban Australian adults,1980—2000. Public Health Nutr 2010;13(5):631—8.

15] Sturm R, Hattori A. Morbid obesity rates continue torise rapidly in the United States. Int J Obes (Lond)2013;37(6):889—91.

16] Keating CL, Peeters A, Swinburn BA, Magliano DJ, MoodieML. Utility-based quality of life associated with overweightand obesity: the australian diabetes, obesity, and lifestylestudy. Obesity (Silver Spring) 2012.

17] Must A, Spadano J, Coakley EH, Field AE, Colditz G, DietzWH. The disease burden associated with overweight andobesity. JAMA 1999;282(16):1523—9.

18] Keating CL, Moodie ML, Bulfone L, Swinburn BA, Steven-son CE, Peeters A. Healthcare utilization and costs inseverely obese subjects before bariatric surgery. Obe-sity 2012;20:2412—9, http://dx.doi.org/10.1038/oby.2012.124.

19] Wong E, Stevenson C, Backholer K, Mannan H, PasupathiK, Hodge A, et al. Adiposity measures as predic-tors of long-term physical disability. Annals Epidemiol2012;22(10):710—6.

20] Swinburn BA, Sacks G, Hall KD, McPherson K, FinegoodDT, Moodie ML, et al. The global obesity pandemic:shaped by global drivers and local environments. Lancet


2011;378(9793):804—14.21] Swinburn B, Egger G. The runaway weight gain train:

too many accelerators, not enough brakes. BMJ2004;329(7468):736—9.

dx.doi.org/10.1016/j.orcp.2015.02.004

http://dx.doi.org/10.1016/j.orcp.2015.02.004

http://refhub.elsevier.com/S1871-403X(15)00024-1/sbref0135






































































































































































































































































































http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4364.0.55.003Explanatory Notes12011-2012?OpenDocument



dx.doi.org/10.1111/obr.12161





















































































































dx.doi.org/10.1038/oby.2012.124

dx.doi.org/10.1038/oby.2012.124



















































































[

[

ARTICLE10

[22] Sjostrom L. Review of the key results from the SwedishObese Subjects (SOS) trial —– a prospective controlledintervention study of bariatric surgery. J Internal Med2013;273(3):219—34.


[23] Keating C, Backholer K, Moodie M, Stevenson C, Peeters A.Differences in the rate of treatment of severe obesity usingbariatric surgery across socio-economic groups. JAMA Surg2015, http://dx.doi.org/10.1001/jamasurg.2014.3180.

Available online at www

ScienceD


24] Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of child-hood and adult obesity in the United States, 2011—2012.JAMA 2014;311(8):806—14.

25] Sperrin M, Marshall AD, Higgins V, Buchan IE, Renehan AG.Slowing down of adult body mass index trend increases


in England: a latent class analysis of cross-sectional sur-veys (1992—2010). Int J Obes (Lond) 2014;38(6):818—24,http://dx.doi.org/10.1038/ijo.2013.161.

.sciencedirect.com

irect

dx.doi.org/10.1016/j.orcp.2015.02.004

































dx.doi.org/10.1001/jamasurg.2014.3180





























dx.doi.org/10.1038/ijo.2013.161

http://www.sciencedirect.com/science/journal/1871403X

Prevalence of class-I, class-II and class-III obesity in Australian adults between 1995 and 2011-12

Documents