The epidemiology of overweight and obesity among Australian children and adolescents, 1995-97

162 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2001 VOL. 25 NO. 2

Submitted: June 2000

Revision invited: November 2000

Accepted: March 2001

Correspondence to:Dr Michael Booth, NSW Centre for the Advancement of Adolescent Health,The Children’s Hospital at Westmead, Locked Bag 4001, Westmead NSW 2145.Fax: (02) 9845-0663; e-mail: [email protected]

Abstract

Objectives: To determine the population

prevalence of overweight and obesity

among Australian children and adolescents,

based on measured body mass index

(BMI). To determine if overweight and

obesity are distributed differentially across

the population of young Australians.

Methods: Data from three independent

surveys were analysed. In each, height and

weight were measured by trained surveyors

using valid, comparable methods. BMI

(kg/m2) was used as the index of adiposity

and recently published international BMI cut-

off values used to categorise each subject

as non-overweight, overweight or obese.

Results: The population prevalence and

distribution of overweight, obesity and

overweight/obesity combined were

generally consistent across datasets. The

ranges of the prevalence of non-overweight,

overweight, obesity and overweight/obesity

combined were 79-81%, 14-16%, 5% and

19-21% (boys) respectively and 76-79%,

16-18%, 5-6% and 21-24% (girls). There

were no consistent relationships between

the prevalence of overweight/obesity and

sex, age or SES. Their prevalence was up

to 4% higher in urban than rural areas

among boys, but there were no differences

between urban and rural girls. The data

suggest a higher prevalence of overweight/

obesity among students from European or

Middle-Eastern cultural backgrounds.

Conclusions: Some 19-23% of Australian

children and adolescents are either

overweight or obese. Although urban/rural,

SES and cultural background differentials

were noted, only the last warrants a

targeted health promotion response.

Implications: Overweight/obesity is a

prevalent health risk factor among

Australian children and adolescents. More

information is needed to understand

whether targeted approaches are required

for specific ethnic groups in addition to

broad, population-based approaches.

(Aust N Z J Public Health 2001; 25: 162-9)

The epidemiology of overweight and obesity among

Australian children and adolescents, 1995-97

Michael L. BoothNSW Centre for the Advancement of Adolescent Health, Department of Paediatricsand Child Health, The University of Sydney at The Children’s Hospital at Westmead,New South Wales

Melissa WakeCentre for Community Child Health, Royal Children’s Hospital, Victoria

Tim ArmstrongNational Centre for Monitoring Cardiovascular Disease, Australian Institute ofHealth and Welfare

Tien CheyEpidemiology Unit, South Western Sydney Area Health Service, New South Wales

Kylie HeskethCentre for Community Child Health, Royal Children’s Hospital, Victoria

Sushma MathurNational Centre for Monitoring Cardiovascular Disease, Australian Institute ofHealth and Welfare

S ignificant physical and psycho-

social health problems are associ-

ated with overweight and obesity in

childhood and adolescence.1,2 Overweight

children and adolescents are also at increased

risk of becoming overweight adults3,4 and of

experiencing the chronic health problems

associated with adult obesity.5

Understanding the epidemiology of over-

weight among children and adolescents is a

critical first step in formulating an appro-

priate public health response. Data on the

prevalence of overweight and obesity allows

description of the magnitude of the problem,

informing decisions about public health pri-

orities and appropriate resource allocation.

Monitoring changes in the prevalence of

overweight and obesity is also important as

public health priorities change.

In addition to monitoring the prevalence of

overweight and obesity among children and

adolescents, it is useful to examine the socio-

demographic distribution of overweight and

obesity in the population. If overweight is found

to be more prevalent among some groups of

children and adolescents, it may be appropri-

ate to allocate health promotion resources dif-

ferentially to those groups and to fashion

interventions to ensure they most closely fit

the characteristics of the target groups.

Although there have been several reports

on the prevalence of overweight and obesity

among Australian children and/or adoles-

cents, all of them have used some type of

relative index of overweight such as 120%

or greater than a ‘standard’ weight-for-height

or 85th or 95th percentiles.6-8 Use of these

relative indices has significant shortcomings,

since it often precludes comparisons across

populations.

Health Promotion

2001 VOL. 25 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 163

Booth and his colleagues,10 using mean values, found no sig-

nificant associations between BMI and school year, urban/rural

place of residence or socio-economic status (SES) among boys.

Among girls, there was a significant association between BMI

and SES, but no other statistically significant associations. A re-

cent report has also shown a significant association between BMI

and ethnicity within a region of Sydney, with students of Medi-

terranean cultural background having significantly greater BMI

than those of Anglo background, who had greater BMI than those

of Asian background.11 Overall, these studies suggest that over-

weight /obesity is greater among urban youth and some cultural

groups, and that overweight/obesity may be inversely associated

with SES. Inferences that could be drawn from these studies have

been limited, however, by the unavailability of absolute criteria

for BMI and sufficiently rigorous statistical analysis.

Absolute criteria for BMI for children and adolescents have

recently been proposed,12 allowing us to report the population

prevalence and distribution of BMI categories for Australian chil-

dren and adolescents for the first time. These absolute BMI cut-

points are applied to three recent datasets: the NSW Schools

Fitness and Physical Activity Survey, 1997; the 1995 National

Nutrition Survey; and the 1997 Health of Young Victorians Study.

We also consider the implications of the findings for the develop-

ment of health promotion interventions aimed at reducing the

prevalence of overweight and obesity among Australian children

and adolescents.

MethodsThe three studies reported on here are briefly introduced and

the methods are summarised in Table 1. The methods of each

study have been described in detail elsewhere (references are pro-

vided in the following text).

The NSW Schools Fitness and Physical Activity Survey 1997

(Study 1) was a state-wide survey (n=5,518) of students in school

years 2, 4, 6, 8 and 10.13,14 Height, weight, waist and hip girths, a

set of measures of health-related fitness and tests of competency

at six fundamental movement skills were administered to students

in Years 4, 6, 8 and 10. Only height and weight were measured

among Year 2 students and high school students also completed a

comprehensive self-report questionnaire.

The 1995 National Nutrition Survey (NNS; Study 2) was a joint

venture between the then Commonwealth Department of Health

and Family Services and the Australian Bureau of Statistics (ABS).15

Health Promotion Overweight and obesity among Australian children and adolescents

Table 1: Summary of the methods of each study.

NSW Schools Fitness and National Nutrition Health of YoungPhysical Activity Survey Survey Victorians Study

When conducted Feb-March 1997 Feb 1995 – March 1996 Sept-Dec 1997

Where conducted NSW (state-wide) Australia (national) Victoria (state-wide)

Sampling method Two-stage stratified random Sub-sample of National Two-stage stratified random samplesample (school & class). Health Survey: a random (school & class). Probability ofProbability of selection of school household survey selection of school proportional toproportional to size of enrolment. size of enrolment.

Height measurement

Scale Portable stadiometer and Portable stadiometer and Invicta (Leicester) portablestretch stature method stretch stature method stadiometer

Precision Nearest 0.1cm Nearest 0.1cm/average Nearest 0.1cmof two measurements

Weight measurement

Scales Tanita Model 1597 Tanita Model 1597 Tanita Model 1597portable digital scales portable digital scales portable digital scales

Precision Nearest 0.1kg Nearest 0.1kg Nearest 0.1kgClothing Light clothes, no shoes Light clothes, no shoes Light clothes, no shoes

Body mass index (BMI)a kg/m2 kg/m2 kg/m2

Socio-demographic measures

School year/age Yrs 2, 4, 6, 8, 10 Decimal years Yrs prep, 1, 2, 3, 4, 5, 6SES IRSDb used to give postcode IRSDb used to give postcode IRSDb used to give postcode

of residence an SES score of residence an SES score of residence an SES score +mother’s years of formal education

Geographic location Urban/ruralc Urban/rurald Urban/rurale

Cultural background English-speaking, European, Not available Oceania, European, Middle East,Middle East, Asian, Otherf Asiang

Notes:(a) BMI categorised as non-overweight, overweight, obese or overweight/obese based on recently developed criteria.12

(b) Australian Bureau of Statistics Index of Relative Socioeconomic Disadvantage (IRSD)18

(c) Urban = Sydney metropolitan area, Newcastle, Wollongong and Blue Mountains. Rural = all others.(d) Urban = postcode located in a State/Territory capital city or centres >100,000 population. Rural = all others19

(e) Based on Australia Post definitions.(f) Based on language spoken most at home.(g) Based on country of birth of responding parent; categorised using ABS classifications.


Table 2: Characteristics of the samples.

NSW Schools Fitness and National Nutrition Survey Health of Young Victorians Physical Activity Survey n=5,518 n=2,822 Study n=2,863

Boys n (%) Girls n (%) Boys n (%) Girls n (%) Boys n (%) Girls n (%)

School yearPrep – – – – 211 (14.6) 216 (15.2)

1 – – – – 178 (12.3) 178 (12.6)

2 606 (20.5) 549 (21.4) – – 220 (15.2) 207 (14.6)3 – – – – 218 (15.1) 215 (15.2)

4 601 (20.4) 532 (20.7) – – 223 (15.4) 214 (15.1)

5 – – – – 181 (12.5) 197 (13.9)6 663 (22.5) 541 (21.1) – – 214 (14.8) 191 (13.5)

8 557 (18.9) 515 (20.1) – – – –

10 524 (17.8) 430 (16.8) – – – –

Age2-5 – – 355 (24.7) 399 (28.9) – –

6-9 – – 395 (27.4) 351 (25.4) – –10-13 – – 381 (26.5) 348 (25.2) – –

14-17 – – 309 (21.5) 284 (20.5) – –

7-15 – – 844 737 – –

Geographical regionUrban 2,059 (69.8) 1,762 (68.6) 504 (59.7) 434 (58.9) 895 (62.0) 863 (61.2)Rural 892 (30.2) 805 (31.4) 340 (40.3) 303 (41.1) 548 (38.0) 546 (38.8)

Cultural backgroundEnglish-speaking 2,407 (81.6) 2,096 (81.7) – – 1,127 (79.8)a 1,097 (78.9)a

European 138 (4.7) 95 (3.7) – – 192 (13.6) 198 (14.2)

Middle East 125 (4.2) 122 (4.8) – – 34 (2.4) 29 (2.1)

Asian 201 (6.8) 188 (7.3) – – 60 (4.2) 66 (4.7)Other 34 (1.2) 27 (1.1) – – – –

Note:(a) Defined as Oceania (including Australia) and Americas.

It was conducted using a sub-sample of the National Health Sur-

vey (NHS), a survey administered to a randomly selected sample

of Australian residents between February 1995 and March 1996.16

The NNS covered urban and rural areas across all States and Terri-

tories of Australia, and included people aged two years or older

who were residents of private dwellings. Only the data collected on

2,819 respondents aged 2-17 years are presented here.

The 1997 Health of Young Victorians Study (Study 3) was a large,

cross-sectional epidemiological study of the health and well-being

of Victorian school children.17 In the primary school sample, data

were collected from 3,104 children (aged 5-13 years) in 24 primary

schools across Victoria. All children had their height and weight

measured and parents provided socio-demographic information.

Recently developed criteria for overweight and obesity among

children and adolescents12 were applied to the data. These criteria

do not rely on percentiles relative to any given population. Rather,

they relate children’s age-adjusted BMI z-score to BMIs of 25

and 30 at age 18 years for males and females respectively, and

should therefore facilitate comparisons across populations and

over time. They provide BMI values for males and females aged

2-18 years in six-month age brackets which can be used as

cutpoints for overweight and obesity. Ages of all subjects were

calculated on the basis of birth dates (self-reported for older sub-

jects or reported by parents or extracted from school records for

younger subjects) and the date of survey response.

Statistical analysisStudies 1 and 3 adjusted for design effects in all analyses and

the data were self-weighted. There were no design effects in Study

2 and the data were weighted to the 1995 Australian population.

Data analyses were carried out using SAS, Sudaan and STATA

software. Number (n) and prevalence (%) of each obesity classi-

fication by demographic factor were tabulated separately by gen-

der. Chi-square tests of signif icant association between

demographic factor and obesity classification were performed

using Sudaan with adjustment for the design effects. Using mul-

tiple logistic regression, the odds of overweight/obesity by SES

category with reference to the lowest quintile were explored after

adjusting for age and demographic characteristics. A test for SES

trend was carried out using SES quintile as a continuous variable

with value 1 (low) to 5 (high). Trends were determined by the

change in deviance (as a chi-square) on one degree of freedom

following the addition of the variable to a regression equation.

ResultsStudy 1: The NSW Schools Fitness and PhysicalActivity Survey, 1997

The response rates for boys and girls in each primary school

year were all greater than 90%, were greater than 80% for Year 8

students and for Year 10 boys, while for Year 10 girls the response

Booth et al. Article


Table 3: NSW Schools Fitness and Physical Activity Survey, 1997: Proportion of boys and girls in each BMI category foreach of the socio-demographic measures.

Non-overweight (%) Overweight (%) Obese (%) Overweight + obese (%)Boys Girls Boys Girls Boys Girls Boys Girls

(n=2,309) (n=1,977) (n=431) (n=410) (n=151) (n=124) (n=582) (n=534)

School year2 81 78 12 15 7 7 19 23

4 79 78 14 17 7 5 21 22

6 80 77 16 19 4 5 20 238 79 82 16 14 6 4 22 19

10 81 80 16 17 3 3 19 20χ2(8)=25.18 χ2(8)=11.02 χ2(4)=2.33 χ2(4)=2.74

p=0.004 p=0.22 p=0.68 p=0.61

SES1 (low) 80 78 14 16 6 7 20 22

2 77 74 17 19 6 7 23 26

3 81 79 14 16 5 6 19 214 79 78 16 20 5 3 21 22

5 (high) 83 84 13 12 5 3 17 16χ2(8)=8.24 χ2(8)=22.52 χ2(4)=5.36 χ2(4)=11.50

p=0.42 p=0.008 p=0.26 p=0.03

NESBEnglish epeaking 81 79 14 16 5 5 19 21

European 67 74 27 22 7 4 33 26Middle Eastern 75 69 15 20 10 12 25 31

Asian 79 85 15 13 7 3 22 16χ2(6)=27.72 χ2(6)=11.58 χ2(3)=16.10 χ2(3)=10.50

p<0.001 p=0.09 p=0.002 p=0.02

Geographic locationUrban 79 79 15 16 6 5 21 21

Rural 82 78 15 17 3 5 18 22χ2(2)=9.5 χ2(2)=0.22 χ2(1)=2.72 χ2(1)=0.22p=0.01 p=0.89 p=0.10 p=0.64

rate was 71%. The majority of cases of non-participation (>70%)

were due to absenteeism on the day of testing rather than refusal

to participate, although it is recognised that absenteeism may be

a passive form of refusal to participate. Characteristics of the sam-

ple are shown in Table 2. Table 3 shows the prevalence of non-

overweight, overweight, obesity and overweight/obesity combined,

for each category of each socio-demographic variable. Table 3

also shows the chi-square statistics for the relationships between

each socio-demographic variable and BMI expressed as two (non-

overweight and overweight/obese) and as three categories (non-

overweight, overweight and obese).

The proportion of girls and boys who were non-overweight,

overweight or obese did not differ by sex when overweight and

obesity were included as separate categories (χ2(2)=0.80, p=0.37)

or when overweight and obesity were combined into a single cat-

egory (χ2(1)=1.97, p=0.38). There was not a significant relation-

ship between overweight/obesity and age. There was no

relationship between SES and overweight for boys, but among

girls the highest SES quintile had a significantly lower propor-

tion of girls who were overweight or overweight/obese and a higher

prevalence of girls with acceptable BMI. The data suggest that

significantly greater proportions of both boys and girls of Euro-

pean or Middle Eastern cultural background were overweight,

obese or overweight/obese than were students of English-speak-

ing or Asian background. Although the prevalence of overweight,

obesity and overweight/obesity was significantly greater among

urban compared with rural boys, the differences between urban

and rural girls were small and not statistically significant.

Table 4 shows, for boys and girls separately, the prevalence and

unadjusted odds ratios for overweight/obesity for quintiles of SES

and the odds ratios adjusted for school year, cultural background

and geographic location. The relationship between SES and over-

weight/obesity was statistically significant for girls, but not boys.

The odds ratio for girls in the highest quintile of girls was signifi-

cantly different from the reference category (lowest quintile of

SES).

Study 2: The National Nutrition SurveyOf the 4,498 2-18 year olds selected from the NHS to partici-

pate in the NNS, 3,711 (83%) accepted. Of those, 3,007 (81%)

participated in the survey. The socio-demographic characteristics

of the sample are shown in Table 2, and Table 5 shows the preva-

lence of under/acceptable weight, overweight, obesity and

overweight/obesity combined, for each category of each socio-

demographic variable. Table 5 also shows the chi-square statis-

tics for the relationships between each socio-demographic



Table 4: Results of logistic regression analyses for each sample. The table shows, for boys and girls separately, theprevalence and fully adjusted odds ratios of overweight/obesity for quintiles of socio-economic status (SES).

NSW Schools Fitness and National Nutrition Survey Health of YoungPhysical Activity Survey Victorians Study

SES prevalence Unadjusted Adjusteda Prevalence Unadjusted Adjustedb Prevalence Unadjusted Adjustedc

quintiles (%) odds ratios odds ratios (%) odds ratios odds ratios (%) odds ratios odds ratios(95% CI) (95% CI) (%) (95% CI) (95% CI) (95% CI)

Boys1 (low) 22.6 1.00 1.00 14.4 1.00 1.00 19.6 1.00 1.00

2 24.9 1.13 (0.87-1.49) 1.21 (0.92-1.59) 19.6 1.18 (0.71-1.65) 1.20 (0.73-1.67) 25.5 1.40 (0.93-2.11) 1.25 (0.82-1.91)

3 20.9 0.91 (0.68-1.20) 1.04 (0.77-1.39) 22.2 1.35 (0.92-1.79) 1.36 (0.92-1.79) 22.1 1.16 (0.79-1.70) 0.87 (0.57-1.31)4 23.4 1.05 (0.79-1.38) 1.08 (0.81-1.44) 18.4 1.08 (0.64-1.51) 1.09 (0.66-1.53) 20.2 1.04 (0.71-1.51) 0.98 (0.66-1.44)

5 (high) 19.0 0.80 (0.60-1.07) 0.81 (0.59-1.09) 20.4 1.17 (0.75-1.58) 1.06 (0.63-1.49) 19.6 1.00 (0.67-1.47) 0.75 (0.50-1.15)

SES trends:Π2

(1) (p-value) 2.67 (ns) 2.43 (ns) 0.13 (ns) 0.001 (ns) 0.16 (ns) 7.32 (p<0.01)

Girls1 (low) 23.1 1.00 1.00 20.4 1.00 1.00 24.0 1.00 1.00

2 30.1 1.43 (1.08-1.90)* 1.45 (1.08-1.93)* 27.0 1.15 (0.73-1.58) 1.17 (0.75-1.59) 24.0 1.00 (0.67-1.49) 0.90 (0.60-1.37)

3 22.8 0.98 (0.74-1.31) 1.03 (0.76-1.39) 22.3 1.00 (0.57-1.42) 1.01 (0.58-1.43) 23.1 0.95 (0.66-1.37) 0.82 (0.55-1.22)4 23.4 1.02 (0.76-1.37) 1.05 (0.77-1.43) 22.9 1.15 (0.75-1.55) 1.15 (0.75-1.55) 21.8 0.88 (0.61-1.27) 0.85 (0.58-1.23)

5 (high) 17.5 0.71 (0.52-0.95)* 0.75 (0.54-1.03) 14.9 0.67 (0.26-1.08) 0.63 (0.21-1.05) 25.5 1.08 (0.75-1.56) 0.92 (0.62-1.38)

SES trends:Π2

(1) (p-value) 9.43 (p<0.01) 5.39 (p<0.01) 3.617 (ns) 3.731 (ns) 0.00 (ns) 1.50 (ns)

Notes:(a) Adjusted for school year (5 categories), cultural background (4 categories) and geographic location (2 categories)(b) Adjusted for age (continuous) and geographic location (2 categories)(c) Adjusted for school year (7 categories), cultural background (4 categories) and geographic location (2 categories)* p<0.05** p<0.01ns = statistically not significant (p>0.05)

variable and BMI expressed as two categories and as three cat-

egories.


overweight or obese did not differ when overweight and obesity

were included as separate categories (χ2(2)=3.99, p=0.14), but

were marginally significantly different when overweight and obes-

ity were combined into a single category (χ2(1)=3.81, p=0.051).

Significantly smaller proportions of boys aged 2-5 years or 6-9

years were overweight, obese or either overweight or obese, but

there was no relationship between overweight and age for girls.

There was no apparent relationship between BMI and SES among

boys, but among girls the prevalence of overweight, obesity and

overweight/obesity was lowest in the highest quintile of SES. The

prevalence of overweight, obesity and overweight/obesity was

similar in urban and rural boys and girls.



and the odds ratios adjusted for age and geographic location. In

the fully adjusted model, the relationship between BMI and SES

was not significant for boys or girls, although the odds ratio for

girls in the highest quintile of SES approached statistical signifi-

cance.

Study 3: The 1997 Health of Young Victorians StudyTwenty-four primary schools participated. The overall student/

parent response rate was 75%. Sixty-one per cent of students

resided in urban areas and 50.5% were male. The characteristics

of the sample are shown in Table 2, and Table 6 shows the preva-

lence of non-overweight, overweight, obesity and overweight/

obesity combined, for each category of each socio-demographic

variable. Table 6 also shows the chi-square statistics for the rela-

tionships between each socio-demographic variable and BMI ex-

pressed as two categories and as three categories.


overweight or obese was not significantly different when over-

weight and obesity were included as separate categories

(χ2(2)=2.5, p=0.29) or when overweight and obesity were com-

bined into a single category (χ2(1)=2.5, p=0.12).

There were no clear patterns by school year or SES in the preva-

lence of non-overweight, overweight, obesity or overweight/obes-

ity for Victorian primary school children. The prevalence of

overweight, obesity and overweight/obesity was significantly

higher for urban boys compared with rural boys, but the differ-

ence between urban and rural girls was not statistically signifi-

cant. The relationship between maternal education and BMI

category was statistically significant for three categories of BMI,

but not for two categories of BMI for both boys and girls.



and the odds ratios adjusted for school year, cultural background

and geographic location. The trend in the prevalence of overweight

and obesity with increasing SES was not statistically significant



Table 5: 1995 National Nutrition Survey: Proportion of boys and girls in each BMI category for each of the socio-demographic measures.


(n=1,162) (n=1,074) (n=207) (n=234) (n=71) (n=74) (n=278) (n=308)

Age2-5 84 77 13 17 3 6 16 23

6-9 86 78 11 15 4 7 15 2210-13 77 76 21 19 3 5 23 25

14-17 77 86 16 11 7 3 23 14χ2(6)=16.44 χ2(6)=6.43 χ2(3)=11.04 χ2(3)=5.72

p=0.01 p=0.38 p=0.01 p=0.13

SES1 (low) 86 80 11 17 3 3 14 20

2 80 73 16 19 4 8 20 27

3 78 78 16 18 6 5 22 224 82 78 16 16 3 7 18 23

5 (high) 80 85 16 12 5 3 20 15χ2(8)=6.15 χ2(8)=13.53 χ2(4)=2.20 χ2(4)=9.94

p=0.63 p=0.10 p=0.70 p=0.04


Rural 81 81 16 15 4 4 19 19χ2(2)=1.26 χ2(2)=0.17 χ2(1)=1.07 χ2(1)=0.15 p=0.53 p=0.92 p=0.30 p=0.70

Table 6: Health of Young Victorians Study, 1997: Proportion of boys and girls in each BMI category for each of thesocio-demographic measures.


(n=1,140) (n=1,084) (n=229) (n=253) (n=76) (n=81) (n=305) (n=334)

School yearPrep 82 80 12 15 6 5 18 20

1 82 75 17 15 2 10 19 252 81 74 15 19 5 6 20 26

3 79 78 13 17 8 5 21 22

4 74 75 20 20 6 5 26 255 76 75 19 19 6 7 24 25

6 79 77 16 19 4 4 21 23χ2(12)=15.8 χ2(12)=10.6 χ2(6)=6.1 χ2(6)=3.1

p=0.20 p=0.56 p=0.42 p=0.79

SES1 (low) 80 76 14 17 5 7 20 24

2 75 76 17 19 9 5 26 24

3 78 77 15 17 7 6 22 234 80 78 16 17 4 5 20 22

5 (high) 80 75 17 21 2 5 20 26χ2(8)=13.5 χ2(8)=4.7 χ2(4)=3.5 χ2(4)=1.1

p=0.10 p=0.79 p=0.47 p=0.89


Rural 85 78 13 18 3 4 16 22 χ2(2)=18.8 χ2(2)=3.1 χ2(1)=16.8 χ2(2)=1.2

p<0.001 p=0.21 p<0.001 p=0.28

Maternal educationYear 10 75 71 17 20 8 8 25 29

Year 11/12 79 77 17 18 4 6 21 23 Tertiary 80 79 15 17 5 4 20 21

χ2(4)=9.3 χ2(4)=10.2 χ2(2)=3.1 χ2(2)=7.5p=0.06 p=0.04 p=0.21 p=0.02

Notes:(a) includes Australia and the Americas



for either boys or girls after adjusting for school year, cultural

background and geographic location.

DiscussionExamination of the data from three large, independent surveys

allows us to look for consistent socio-demographic trends in the

prevalence of overweight and obesity and to determine if all three

provide similar estimates of the overall prevalence of overweight

and obesity. The great advantage of having several such datasets

is that findings which are consistent across all three datasets are

likely to represent real phenomena, not artefact or spurious

results. However, unless surveys are specifically designed for

direct comparison, it is inevitable that methodological differences

will arise, which should give us pause in making comparisons

between sets of findings. Although the surveys reported here have

a number of important features in common, there are also sub-

stantial differences.

All three surveys measured height and weight using reliable

and comparable techniques, all selected subjects at random and

had acceptable response rates. However, the surveys employed

different sampling frames, were conducted in different geographi-

cal regions, included children and adolescents of different age

ranges and administered the measures in different settings (homes

and schools). Each of these methodological differences may, plau-

sibly, have an effect on the apparent prevalence and demographic

distribution of overweight and obesity, but we are unable to make

judgements about the direction and magnitude of the effects of

these differences.

With regard to age (or its proxy, school year), no clear relation-

ships are evident for either boys or girls. However, because the

international cutpoints are standardised for age and gender, we

would not expect to find differences between boys and girls or by

age unless the shapes of our population curves are substantially

different from those of the six pooled reference populations from

which the international cutpoints were derived. A very large

representative sample (>10,000 children) would be required to

develop reliable Australian BMI curves and compare them with

the international reference set of curves, to determine whether

patterns of BMI do differ by age and gender in unexpected ways

from the international reference set.

Among girls, there are significant relationships between SES

and BMI category in only one of the datasets (NSW). It appears

that if there is an association between SES and BMI, it is only of

modest strength. In the NSW data, the lowest odds ratio is for the

highest SES quintile and there appears to be little difference among

the four lower SES quintiles. This pattern of results may be due to

the different age ranges of the surveys. A close examination of

the NSW and NNS data by SES (see Tables 3 and 5) suggests that

there is little difference in the prevalence of overweight and obes-

ity among the younger survey respondents, but that the preva-

lence of overweight and obesity falls among adolescents (aged

approximately 14-17 years). In this case, it is not surprising that

no association was apparent in the Victorian data, which were

limited to primary school children. It may be that the “culture of

thinness” has a more potent influence on adolescent girls of higher,

compared with lower, SES. These findings appear to be some-

what inconsistent with the findings for maternal education, which

suggest a greater prevalence of overweight/obesity among young

people of lower SES. This apparent discrepancy may be due to

the fact that a mother’s years of formal education and postcode of

residence provide different indices of SES, between which there

is only poor to moderate agreement. The findings on the relation

between SES and BMI should be viewed with some caution until

this issue is resolved.

There appears to be a trend towards a greater prevalence of

overweight/obesity among boys resident in urban areas, but the

observed differences may be due to the confounding effect of

cultural background, as young people of Middle Eastern and Eu-

ropean background are more likely to live in urban areas. If that

was the case, however, one would also expect to see a higher preva-

lence of overweight and obesity among urban girls, which is not

evident in the data. Interpretation of the relationship between BMI

and cultural background is made difficult by the small numbers

in some of the groups and the different definitions used in the

different surveys. Nevertheless, the findings do suggest that there

are very substantial differences between children and adolescents

from different cultural backgrounds, a result warranting further

investigation.

Each dataset has its specific shortcomings. For the NSW sur-

vey, students were sampled from every second year and it is not

known if the students of the same age as those sampled, but in

different school years, may have differed systematically from those

sampled. The NSW and Victorian surveys were limited to stu-

dents attending school. While almost all children of primary school

and early high school age are likely to attend school, early school

leaving starts in about Year 9. Consequently, it cannot be assumed

that students attending Year 10 are necessarily representative of

the entire population of 15-year-olds.

Although the response rates were reasonably high for all of the

surveys, it is still possible that there is a non-response bias in the

data. It is quite plausible that children and adolescents who are

significantly over- or underweight are more likely to deliberately

avoid an anthropometric assessment. It would be helpful in this

regard if, in future studies, efforts were made to collect data on

all students in a number of schools to determine if initial non-

responders are systematically different from those who agree to

participate at the first invitation. Despite these limitations, the

three studies reported here were generally methodologically sound

and the prevalences are likely to be reasonably representative

estimates. Furthermore, the findings are generally consistent with

the findings of earlier Australian studies which used different cri-

teria of overweight.6-9

Effective interventions and policy initiatives intended to

address overweight and obesity among young people are sorely

needed. Results of these three recent, large representative studies

highlight the extent of the problem across all age, gender

and socio-demographic categories examined. They suggest that



initiatives must be broad: it is not appropriate to differentially

target any one group over another. Better information about spe-

cific ethnic groups may change this recommendation for a

minority, but will not alter the need for broad interventions for

the majority of Australian children and youth.

AcknowledgementsThe study was supported, in part, by grants from the NSW

Department of Education and Training, the NSW Department of

Health, and the National Professional Development Program.

References1. Dietz WH. Prevention of childhood obesity. Pediatr Clin North Am

1986;33:823-33.2. French SA, Story M, Perry CL. Self-esteem and obesity in children and ado-

lescents: A literature review. Obes Res 1995;3: 479-90.3. Must A. Morbidity and mortality associated with elevated body weight in

children and adolescents. Am J Clin Nutr 1996;63(suppl):445S-7S.4. Srinivasan SR, Bao W, Wattigney WA, Berenson G. Adolescent overweight is

associated with adult overweight and related multiple cardiovascular risk fac-tors: The Bogalusa Heart Study. Metabolism 1996;45:235-40.

5. Must A, Jacques P, Dallal G, Bajema C, Dietz W. Long-term morbidity andmortality of overweight adolescents: A follow-up of the Harvard Growth Studyof 1922 to 1935. New Engl J Med 1992;327:1350-55.

6. Court JM, Dunlop M, Reynolds M, Russell J, Griffiths L. Growth and devel-opment of fat in adolescent school children in Victoria. Part 1. Normal growthvalues and prevalence of obesity. Aust Paediatr J 1976;12:296-304.

7. Simons LA, Anderson N, Simons J, Whish P. Health attitudes and knowledge,and coronary risk factors in high-school children, Sydney and Inverell. MedJ Aust 1982;2:178-83.

8. Harvey PWJ, Althaus M-M. The distribution of body mass index in Austral-ian children aged 7-15 years. Aust J Nutr Diet 1993;50(4):151-3.

9. Court JM, Dunlop M, Reynolds M, Russell J, Griffiths L. Growth and devel-opment of fat in adolescent school children in Victoria. Part 2. Influence ofethnic, geographic and socio-economic factors. Aust Paediatr J 1976;12:305-12.

10. Booth ML, Macaskill P, Lazarus R, Baur LA. Sociodemographic distributionof measures of body fatness among children and adolescents in New SouthWales, Australia. Int J Obes 1999;23:456-72.

11. Lynch J, Wang XL, Wilcken DEL. Body mass index in Australian children:recent changes and relevance of ethnicity. Arch Dis Child 2000;82:16-20.

12. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard defini-tion for child overweight and obesity worldwide: international survey.Br Med J 2000;320:1240-3.

13. Booth ML, Macaskill P, McLellan L, Phongsavan P, et al. NSW Schools Fit-ness and Physical Activity Survey, 1997. Sydney: NSW Department of SchoolEducation; 1997.

14. Booth ML, Macaskill P, Phongsavan P, McLellan, L, Okely T. Methods of theNSW Schools Fitness and Physical Activity Survey, 1997. J Sci Med Sport1998;1(2):111-24.

15. Australian Bureau of Statistics. National Nutrition Survey: User’s Guide, 1995.Canberra: AGPS; 1998. ABS Cat no.: 4801.0.

16. Australian Bureau of Statistics. National Health Survey: User’s Guide, 1995.Canberra: AGPS; 1996. ABS Cat no.: 4363.0

17. Lazarus R, Wake M, Hesketh K, Waters E. Change in body mass index inAustralian primary school children, 1985-1997. Int J Obes 2000;24(6):679-84.

18. Australian Bureau of Statistics. Information Paper: 1991 Census Socio-eco-nomic Indices for Areas. Canberra: AGPS; 1993. ABS Cat no.: 2912.0.

19. Department of Primary Industries and Energy. Rural, Remote and Metropoli-tan Areas Classification: 1991 Census Edition. Canberra: AGPS; 1994.


Documents

The epidemiology of overweight and obesity among Australian children and adolescents, 1995-97