• Open Access

South Australian four-year-old Aboriginal children: residence and socioeconomic status influence weight


Dr Nicola Spurrier, Health Promotion Branch, Citi Centre, 11 Hindmarsh Square, Adelaide, SA 5000; e-mail: nicola.spurrier@health.sa.gov.au


Objective: Obesity rates have increased in children in Australia in the past 15 years. However, there is little available population data describing rates of overweight and obesity in Aboriginal children.

Methods: Anthropometric data of four-year-old children (n=11,859) were collected by trained nurses at routine statewide preschool health checks during 2009. Weight status (underweight, healthy weight, overweight and obese) was determined using age and gender specific International Obesity Task Force (IOTF) cut-points.

Results: There were 337 Aboriginal children (3%) in the study population. Aboriginal children had significantly higher rates of overweight and obesity compared to non-Aboriginal children (28% compared to 18% respectively, χ2p=0.0001). A statistically significant association between BMIz score and identifying as Aboriginal remained after controlling for rural/urban residence and socioeconomic status using multiple regression analyses.

Conclusions: Aboriginal children have higher rates of overweight and obesity compared to their non-Aboriginal peers by the time they are four years of age. Aboriginal children have higher BMIz scores compared to non-Aboriginal children after controlling for rural/urban residence and socioeconomic status.

Implications: A significant investment is required to optimise the health of Aboriginal women before pregnancy and throughout pregnancy. A rethink may be necessary in the approach to dietary management and catch-up growth of Aboriginal children of low birth weight or having growth failure in early childhood.

The rates of obesity in Australian children have been reliably collected and reported from a number of sources. The rate of overweight and obesity in children was fairly stable at approximately 10% until the 1970s and has shown a steep increase since then.1 The 2007 Australian National Children's Nutrition and Physical Activity Survey reported that around 23% of Australian children aged 2–16 are either overweight or obese.2 To assess possible ethnic differences, some studies have included sufficient numbers of culturally and linguistically diverse children to allow valid comparisons with Caucasian children.3,4

However, there are few publications to date providing contemporary population-wide anthropometric data for Aboriginal children. Kagawa et al. reported body composition and anthropometric data from a convenience sample of 70 Aboriginal and 75 non-Aboriginal children living in an urban setting.5 They reported significantly higher rates of overweight and obesity in Aboriginal compared to non-Aboriginal girls but with no difference comparing the two groups of boys. However, the rates of overweight and obesity in the non-Aboriginal group were considerably less (7.7% and 0% respectively) than Australian population data, suggesting that the study sample was non-representative. A small study reported height and weight data of 129 Aboriginal children living in an outer metropolitan suburb of Brisbane as part of the Aboriginal and Torres Strait Islander Child Health Check.6 These analyses suggested that the rates of obesity in Aboriginal children aged 5–14 years may be considerably higher than those of non-Aboriginal children.6 Wake et al. published population data collected as part of the Longitudinal Study of Australian Children and this included 181, 4–5 year old Aboriginal children. Estimates for overweight and obesity were 17% and 8% respectively, significantly higher than that of their non-Aboriginal peers.7

The 2004/05 National Aboriginal and Torres Strait Islander Health Survey (NATSIHS), using self-reported weight and height, reported that 29% of Aboriginal adults (aged 15 years and older) were overweight and 31% were obese.8 After adjusting for age differences, Aboriginal women were 2.2 times and Aboriginal men 1.6 times more likely to be obese, compared to non-Aboriginal women and men respectively.8 Indigenous populations in the US, New Zealand and Canada have also reported higher rates of obesity than the dominant non-Indigenous population.9–12 In addition, there are many reports from developing countries showing increasing rates of obesity particularly in urban centres and associated with relative affluence.13

Many geographical, historical, social and economic factors are likely to contribute to overweight and obesity among Indigenous people in Australia and other countries. For example, previous Australian social policy, such as the so-called Protection laws, exempted employers of Aboriginal people from paying a wage if food was provided.15 As a result, many older Aboriginal adults had a subsistence diet of sugar, white flour, tea and lard. Through newly described epigenetic mechanisms, it is possible that this nutrient-poor, high-carbohydrate, low-protein diet could have a negative impact not only on individuals but also on the health of future generations.16 This is a strong argument to ensure that the current generation of Aboriginal women receive optimum nutrition during the child-bearing years and pregnancy, and that a focus on good nutrition continues throughout childhood as prevention against the future burden of chronic disease.

Historically, failure to thrive and malnutrition have characterised the nutritional problems experienced by Aboriginal children.14,17 While there have been some improvements, failure to thrive in the first 2–3 years of life remains a significant public health issue in many remote Aboriginal communities (D. Scrimgeour, personal communication). Given the increases in obesity and chronic disease documented in Aboriginal adults, and the increase in obesity in non-Aboriginal Australian children, it is important to examine whether obesity rates have also increased in Aboriginal children. In South Australia, more than 50% of Aboriginal children live in an urban setting. Epidemiological studies need to include sufficient numbers of children living in urban, rural and remote settings. In addition, differentiating between underweight and normal weight is important given the ongoing concern about ‘failure to thrive’ in some remote communities.

This study aimed to determine the prevalence of underweight, healthy weight, overweight and obesity in four-year-old Aboriginal children living in South Australia in 2009 and to compare this to the prevalence in non-Aboriginal children. A secondary aim was to describe differences in children's weight status between rural, remote and urban settings. In this paper, the term Aboriginal refers to children considered by their parents to be Aboriginal as well as those considered to be Torres Strait Islander.


The preschool health assessment is conducted by Child and Family Health Service (CaFHS) staff in preschools and kindergartens throughout South Australia on behalf of the Women's and Children's Health Network (WCHN). This assessment previously aimed to cover the whole preschool population in SA. Due to changing circumstances and service delivery models in the past 10 years, approximately 65% of children now receive an assessment in the preschool setting. Children in middle to low socioeconomic areas are more likely to be offered the service. A database containing the results of this population-wide preschool assessment has been maintained for more than 15 years, and population estimates of overweight and obesity rates using this database have previously been published.18

In 2009, 12,511 children underwent a preschool health assessment by trained CaFHS community health nurses. The majority of assessments were conducted either in a kindergarten setting (most) or a CaFHS centre. All parents gave written permission for their child to participate and this included consent for their child's data to be stored on the CaFHS data base. Nurses followed a standard protocol to measure children's height with a fixed tape to the nearest 0.01 m and children's weight to the nearest 0.1 kg on digital scales with the child in their underwear and with their shoes removed.

Ethics approval to conduct this study was given by the Women's and Children's Hospital Research Ethics Committee and the Aboriginal Research Ethics Committee of South Australia.

De-identified data was extracted from the CaFHS database for the 2009 year. The 2009 data set was chosen because: 1) it was the most recent; and 2) data regarding Aboriginality in earlier years was considered less reliable. Changes were made during 2008 to improve the capture of Aboriginal status in the database.

Data on 12,160 children were available. Data extracted included children's birth date, date of preschool assessment, gender, height, weight and postcode. The available data were assessed for outliers, data entry errors and missing data. In particular, data from children aged <3 or >6 years were removed and data from children with BMI<10 or >40 were removed as this was considered likely to represent a data entry error. Entry errors were considered random and mostly consisted of missing or extra digits on recorded weight or height. This resulted in removal of 301 cases leaving a sample of 11,859 children (95% of available subjects).

Data was analysed using STATA 10. Body mass index was calculated as weight in kilograms divided by height in metres squared (weight (kg)/(height (m))2). To standardise for age and gender, BMIz scores were calculated using the 1990 British Growth Reference Data and weight status (underweight, healthy weight, overweight and obese) determined using age and gender specific International Obesity Task Force (IOTF) cut-points.19 While direct measures of socioeconomic status were not available in this dataset, most children had a postcode recorded. Postcodes were matched to the Index of Relative Socioeconomic Disadvantage (IRSD), an index derived by the Australian Bureau of Statistics from population census data.20 The IRSD provides a score for each postcode area by summarising characteristics of the population including low income, low educational attainment, high unemployment and unskilled jobs. The index has a base of 1,000 for Australia; scores above 1,000 indicate relative advantage, those below 1,000, relative disadvantage.

Descriptive statistics were obtained and Chi-square tests were used to compare differences between Aboriginal children and non-Aboriginal children categorised as underweight, normal weight, overweight and obese. Chi-square tests were also used to assess rural/remote and urban and gender differences. Because the measure of socioeconomic status (the IRSD score) is a continuous measure, analysis of variance was used to assess differences in IRSD between categories of weight. Finally, multiple linear regression analysis was used to control for potential confounding variables in the association between weight (defined in this instance as BMIz score) and Aboriginality. Interaction between Aboriginality and area of residence was also assessed to see whether the effect of Aboriginality on BMIz was significantly modified by rural/remote versus urban residence.


Aboriginal children represented 3% of the study population (337 children in total). This is slightly higher than the proportion of Aboriginal people in the total South Australian population (1.8%) reflecting the younger age profile of the Aboriginal population.21 There were approximately equal numbers of boys (51%) and girls (49%). Within the Aboriginal group, there were slightly fewer boys (49%) than girls (51%) and within the non-Aboriginal group there was slightly more boys (52%) than girls (48%) (χ2p=0.4). Children's ages ranged from 3.5 to 6 years with an average age of 4.8 years. Table 1 gives mean values for anthropometric measures. The overall rates of overweight (14.2%) and obesity (4.4%) are similar to those reported in other Australian studies.7,18

Table 1.  Descriptive characteristics of total sample (n=11,859).
Variable Number (%) or Mean ± SD (range)
GenderMale6,080 (51.3)
 Female5,779 (48.7)
 Aboriginal337 (3.1)
 Non-Aboriginal10,668 (96.9)
Age (years) 4.76 ± 0.24 (3.52–5.99)
AnthropometryWeight (kg)19.2 ± 2.88 (11.0–51.8)
 Height (m)1.08 ± 0.05 (0.75–1.26)
 BMI (m/kg2)16.16 ± 1.65 (10.17–37.33)
 BMIz scores0.32 ± 1.06 (−7.53 – 6.33)
 BMI category 
  Underweight716 (6.04)
  Healthy weight8,941 (75.4)
  Overweight1,685 (14.2)
  Obese517 (4.4)
Socioeconomic statusIndex of Relative Social Disadvantage978.8 ± 71.9 (726–1128)

Figure 1 shows differences in weight category comparing Aboriginal to non-Aboriginal children. The Chi-square test comparing all the groups was significant (p=0.0001). Comparing population finite confidence intervals, the statistically significant differences were between the proportion of Aboriginal children compared to non-Aboriginal children in the normal weight category (lower proportion of Aboriginal children) and the obese category (higher proportion of Aboriginal children). The proportion of Aboriginal children in the underweight category was slightly higher than non-Aboriginal children but did not reach statistical significance. Similarly, there was a trend towards a higher proportion of Aboriginal children being in the overweight category. Overall, 28% of Aboriginal children were in the overweight or obese category compared to 18% of non-Aboriginal children.

Figure 1.

Distribution of weight categories comparing Aboriginal and non-Aboriginal children.

In terms of gender, significantly more girls than boys were categorised as overweight or obese (χ2p=0.0001). Subgroup analyses showed that Aboriginal girls had the highest rates of obesity (12.9%) (Table 2). Children living in rural areas were more likely to be overweight or obese than those in metropolitan areas (χ2p=0.02). Subgroup analyses showed that there was a higher proportion of Aboriginal children living rurally than non-Aboriginal children and that rural dwelling Aboriginal children had the highest rates of obesity (11.9%) (Table 3).

Table 2.  Distribution of weight categories by gender.
BMI categoryAboriginal n (%)Non-Aboriginal n (%)
Underweight12 (7.2)11 (6.4)319 (5.8)311 (6.0)
Healthy weight111 (66.9)109 (63.7)4,283 (77.8)3,820 (73.7)
Overweight29 (17.5)29 (16.9)729 (13.2)786 (15.2)
Obese14 (8.4)22 (12.9)175 (3.2)265 (5.1)
 χ2p=0.6 χ2p=0.0001 
TOTAL337 10,688 
Table 3.  Distribution of weight categories by place of residence.
BMI categoryAboriginal n (%)Non-Aboriginal n (%)
Underweight12 (7.1)11 (6.6)427 (6.2)199 (5.3)
Healthy weight116 (69.1)104 (61.9)5,257 (76.1)2,823 (75.5)
Overweight24 (14.3)33 (19.6)963 (13.9)543 (14.5)
Obese16 (9.5)20 (11.9)266 (3.9)172 (4.6)
 χ2p=0.5 χ2 p=0.07 
Total336 10,650 

Differences in IRSD score were compared between weight categories using analysis of variance (Table 4). With data from Aboriginal and non-Aboriginal children combined, significant differences in IRSD score were found between weight category; children who were categorised as obese had significantly lower IRSD than children who were categorised underweight, normal weight and overweight (Table 3). When the data is presented separately by Aboriginality, the table also shows that Aboriginal children had statistically significantly lower IRSD scores (mean = 928 ± 3.5, 95% CI 921–935) compared to non-Aboriginal children (mean = 980 ± 0.7 95% CI 979–981). In addition, whilst the numbers are small, the data suggests that a different pattern exists for Aboriginal children compared to non-Aboriginal children. Whilst non-Aboriginal children who are obese have lower IRSD scores, Aboriginal children who are obese tend to have relatively higher scores. In addition, Aboriginal children who are underweight have the lowest IRSD scores of all, whereas for non-Aboriginal children, children who are underweight have higher IRSD scores than children who are obese.

Table 4.  Distribution of weight categories by Index of Relative Social Disadvantage (IRSD).
   IRSD score (mean ± SD) 
 UnderweightHealthy weightOverweightObesep value
  1. a,b Superscripts indicate which categories show a statistically significant difference using Bonferroni correction: same letter indicates no difference, different letter indicates a difference

All groups combined (n=11813)979 ± 73.3a980 ± 72.0a976 ± 71.2a966 ± 69.0b0.0001
Aboriginal children only (n=336)914 ± 51.7a923 ± 67.1a941 ± 63.2a948 ± 56.9a0.04
Non-Aboriginal children only (n=10650)981 ± 73.2a981 ± 71.6ab976 ± 70.6ac967 ± 69.4bc0.00001

A multiple regression analysis was undertaken to assess the independent effect of Aboriginality on children's body mass index. This was because rural residence and IRSD score were associated with both children's weight category and Aboriginality and thus were potential confounding variables. Controlling for place of residence and IRSD score, Aboriginal children still had significantly higher BMIz scores compared to non-Aboriginal children (Table 5). The interaction term calculated between place of residence and Aboriginality did not reach statistical significance when entered into the model (results not shown).

Table 5.  Multiple regression model showing the association between children's BMIz scores and Aboriginality, controlling for place of residence and Index of Relative Socioeconomic Disadvantage (IRSD).
BMIzB ± SEp value
Aboriginal (compared to non-Aboriginal)0.25 ± 0.060.0001
Rural (compared to metro)0.09 ± 0.020.0001
IRSD-0.0004 ± 0.140.008


This study provides unique population data about the level of overweight and obesity in 4–5-year-old Aboriginal children. Young Aboriginal children living in South Australia have 2–3 times higher rates of obesity compared to non-Aboriginal children. While difference in the rates of overweight did not reach statistical significance based on confidence interval criteria, the trend is consistent with the difference in obesity rates. Combining these two categories, 28% of Aboriginal children were categorised as overweight or obese compared to 18% of non-Aboriginal children. These differences resulted in 10% fewer 4-year-old Aboriginal children being in the normal weight category than non-Aboriginal children (a statistically significant difference).

Children living in rural areas of Australia have higher rates of obesity and additionally there is a significant association between socioeconomic status and childhood obesity.7 Aboriginal children are both more likely to live rurally and be socially disadvantaged, thus it is important to control for these confounding variables. Multivariable analyses demonstrated that the significant association between Aboriginality and BMIz score remained after controlling for place of residence and socioeconomic status.

Our results are in contrast to the past where malnutrition and failure to thrive has been the experience of many Aboriginal children. One factor that could have contributed to this finding is that our sample may have been biased towards Aboriginal children living in urban and rural areas rather than remote locations. However, our sample is likely to be representative because it specifically includes children living in remote areas of the state including Yalata, Oak Valley and Oodnadata. The one group not represented in this study are small communities in the Anangu Pitjantjatjara Yankunytjatjara (APY) lands of South Australia because these communities are not involved in preschool screening by the Child and Family Health Service. Using population projections from Aboriginal births, we estimate that there were approximately 650, 4-year-old Aboriginal children living in South Australia in 2009. This suggests that our study includes approximately 50% of South Australian 4-year-old Aboriginal children.

The primary nutrition problem changing from one of ‘failure to thrive’ to obesity is likely to be a true reflection of the fact that Aboriginal people are experiencing the same nutrition transition experienced by people living in developing nations.13 This premise is supported by the results of our study. While for non-Aboriginal children there is a trend for increasing body mass with increasing disadvantage, for Aboriginal children the trend is in the opposite direction. Aboriginal children with lower levels of disadvantage have higher rates of obesity.

There is considerable evidence supporting the hypothesis that early life events strongly influence subsequent growth and development of chronic health conditions.22 The fetal origins (or ‘Barker’) hypothesis, suggests that when nutritional intake of the fetus is compromised, physiological and metabolic programming is altered and results in an increased risk of diabetes, heart disease and obesity later in life.22 Aboriginal infants are significantly more likely to be born small for gestational age, reflecting a less healthy in utero environment secondary to young maternal age, inadequate nutrition during pregnancy and cigarette smoking.23 Such infants, according to the ‘Barker hypothesis’, may be at greater risk of childhood obesity, adult obesity and chronic disease later in life given a similar postnatal environment.24 As noted previously, past Australian social policy during the periods of Protectionism and Assimilation may continue to have profound effects on the health of Aboriginal Australians through recently described epigenetic mechanisms.16

Importantly, obesity does not imply adequate nor nutritionally sound dietary intake. Whilst this study did not aim to address this question, other studies have shown that Aboriginal Australians have reduced access to foods rich in micronutrients and antioxidants such as fresh fruits and vegetables.25 This may be because of lack of access in remote communities where distance and refrigeration make fresh fruit and vegetables very expensive. In addition, many Aboriginal families experience intergenerational poverty and unemployment and have to deal with unstable and low family incomes.26 In such settings, parents may rely on energy dense, nutritionally poor foods to feed themselves and their children. Even if Aboriginal parents were to receive targeted health education on improving children's physical activity and nutrition, this is unlikely to overcome the environmental, economic and social barriers faced by many Aboriginal families and more far-reaching economic improvements are necessary.

There are a number of important implications from this study. The rates of chronic disease experienced by Aboriginal adults is likely to remain high despite planned improvements in health service access through funding programs such as ‘Closing the Gap’. As the current cohort of Aboriginal children move into adulthood, tracking of obesity and early endocrine programming have the potential to result in even higher Aboriginal adult morbidity and mortality through chronic disease in the future. There is a pressing need to look beyond optimising chronic disease management in Aboriginal adults to investing in primary prevention throughout pregnancy and early childhood.

Research is urgently needed to better understand the pathways leading to childhood obesity in Aboriginal children. Accurate longitudinal data is required to assess the relationships between birth weight, catch-up growth, early childhood nutrition and physical activity and subsequent obesity. This may be achieved through increasing the number of annual Aboriginal child health checks undertaken but only if systems are developed to allow capture of this data in an ethical and meaningful way. As has occurred in other settings such as neonatal nurseries, it may be timely to rethink the approach to dietary supplementation and catch-up growth of Aboriginal children of low birth weight or having failure to thrive to ensure optimal and not excessive growth.27 Supporting programs to maintain high rates of breast feeding in Aboriginal communities may be a key intervention.

State and Territory Governments need to be aware of the high rates of obesity in Aboriginal children and ensure that all obesity prevention programs and policies are culturally inclusive and appropriate for Aboriginal children and families. To ensure that the difference in obesity rates does not widen further, such programs need to include a strong Indigenous voice. Similarly, some degree of planning of obesity services for adults and children with significant obesity has commenced in a number of Australian States. While this area of clinical need is under-resourced, particularly with respect to children's services, such services in the public sector will need to be accessible and responsive to Aboriginal children and their families.28

This study has a number of strengths. Because data was collected as part of statewide screening in preschools, the results are likely representative of the South Australian population. Data was collected for both Aboriginal and non-Aboriginal children using the same methodology and so are directly comparable. Anthropometric data was collected directly by trained and experienced nursing staff and is not subject to the reporting bias inherent in self-reported height and weight. Using the most recent international definitions of underweight, we have separately reported rates of underweight. These numbers are usually included in normal weight estimates. This is particularly important for Aboriginal children, where rates of under-nutrition particularly in the preschool years may still be high. This has allowed us to describe the different pattern of obesity by socioeconomic status comparing Aboriginal and non-Aboriginal children.

In terms of limitations, the most significant is that children from the APY lands are not represented in this data set. In addition we considered that there may have been a difference in response rates between Aboriginal children (∼50%) and non-Aboriginal children (∼65%). While overall there is some targeting of health checks towards children in middle and low income areas, it is possible that Aboriginal children who are more socioeconomically advantaged are more likely to have a health check. Because our data shows that there is a reverse gradient between obesity and socioeconomic status by Aboriginal status, these effects should cancel out. The study reports cross-sectional data and therefore cannot measure change across generations. Finally, the data set does not contain birth weights for children at this stage and thus further exploration of the ‘Barker’ or early origins hypothesis is not possible.


Given the higher rates of obesity in Aboriginal adults compared to non-Aboriginal adults, and the overall increases in obesity in Australian children, it is not surprising that our study reports high rates of overweight and obesity in Aboriginal children. This information has important implications for policy and practice. There has been a justifiably strong focus on improving chronic disease management and outcomes in Aboriginal communities, but we believe that a similar effort is required in early life. Maximising the health and nutrition of young women prior to pregnancy and during pregnancy and ensuring that infants and young children are breast fed and receive a nutritionally sound early diet will be important in realising intergenerational improvements in Aboriginal health and wellbeing. In addition, all current and planned childhood obesity prevention and management programs need to be culturally accessible and responsive to Aboriginal families.


The authors thank April Lawrie-Smith, Director, Aboriginal Health Branch, SA Department of Health and Karen Glover, Director, Aboriginal Health, Women's and Children's Health Network for their valuable comments during the development of this paper.