Data sources
The study was based on the secondary analysis of the 2018–2019 National Aboriginal and Torres Strait Islander Health Survey (NATSIHS) data. The NATSIHS was designed to collect a wide range of information about the health of Indigenous Australians, including the prevalence of health conditions (e.g. type 2 diabetes, hypertension, heart diseases), the prevalence of health risk factors (such as smoking and vaping, alcohol consumption and physical activity) and social determinants of health. The NATSIHS was funded by the Australian Government Departments of Health and Prime Minister and Cabinet and implemented by the Australian Bureau of Statistics (ABS) [9]. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.
Study setting and sampling
The NATSIHS was conducted between July 2018 and April 2019. Two samples were used: a community sample and a non-community sample. The community sample included a random selection of discrete Indigenous communities and associated outstations from the Dwelling Register for Aboriginal and Torres Strait Islander Communities. The non-community sample involved multistage area sampling of private dwellings outside Indigenous communities. Mesh blocks with Indigenous households from the 2016 census were identified. Dwellings in each mesh block were randomly selected. In non-remote areas, up to two adults aged 18 years or older were randomly selected from both the community and non-community samples, while in remote areas, up to one adult was randomly selected [9]. For this study, a total of weighted 4864 adult samples (18 years or older) were included.
Cardiometabolic disorders
The main outcome variables were type 2 diabetes, hypertension, high cholesterol and heart disease, measured through self-reported questionnaires. Respondents indicated whether a doctor or nurse had ever diagnosed them with these conditions, and for those with chronic conditions, respondents were additionally asked if the condition was long-term, defined as present at the time of the survey and expected to last 6 months or more [18]. The classification of health conditions in the NATSIHS followed the International Classification of Diseases, 10th Revision (ICD-10) [18]. For this study, the condition of heart diseases included angina, heart attack, heart failure and other heart diseases, based on the ICD-10 criteria.
Anthropometric measurements
Physical measurements, including height, weight and waist circumference (WC), were obtained from respondents aged 2 years and above [18]. Pregnant women were excluded from measurements. Individuals aged 2 years and older were invited in the NATSIHS to voluntarily measure their waist using a tape measure with a maximum length of 150 cm [18]. For this study, central obesity was defined as a waist circumference [WC] of ≥ 102 cm for males and ≥ 88 cm for females [19]. Additionally, the NATSIHS recorded BMI using participants’ voluntarily provided height and weight measurements [18]. BMI was categorised into three groups: normal weight (BMI: 18.5–24.9 kg/m2), overweight (BMI: 25–29.9 kg/m2) and obese (BMI ≥ 30.0 kg/m2) [19].
The term ‘normal weight’ is widely used in the literature to describe BMI values between 18.5 and 24.9 kg/m2. While we used for consistency with established definitions, it is important to note that these classifications may not fully account for the diversity in body composition across populations. For this study, normal-weight central obesity refers to individuals who fall within a typical healthy BMI range (18.5–24.9 kg/m2) but with an elevated WC that meets the thresholds for central obesity.
Other study variables
Based on the review of the literature, we have also included other covariates broadly classified as sociodemographic factors, health risk factors and geographic factors. Sociodemographic factors included age (grouped as ‘18‒29 years’, ‘30‒34 years’ or ‘45 + years’), sex (grouped as ‘male’ or ‘female’), highest educational attainment (grouped as ‘did not complete year 12’, ‘completed year 12’, ‘trade certification or diploma’ or ‘tertiary education’), individual income (grouped as ‘average or more income’ or ‘below average’) and marital status (‘married’ or ‘not married’). We used the 2021 census median individual weekly income (AUD$540) for Indigenous populations to dichotomise income to ‘average or more’ and ‘below average’.
Health risk factors encompassed physical inactivity, low fruit and vegetable consumption, smoking, sugar/sweet drink intake and alcohol consumption. Physical inactivity was defined according to the Australian Department of Health 2014 Physical Activity and Sedentary Behaviour guidelines [20]. Physical activity in this survey was measured using self-reported Recent Physical Activity Questionnaire (RPAQ) [18]. Individuals aged 18–64 years were considered to have met the 2014 Guidelines if, in the past week, they engaged in activities such as walking, moderate or vigorous physical activity on at least 5 days, accumulated 150 min or more (with vigorous activity counted as double) and performed strength or toning exercises on at least 2 days, excluding workplace activity. For individuals aged 65 and over, meeting the guidelines required engaging in activities like walking, moderate or vigorous physical activity every day and accumulating at least 30 min of activity on at least 5 days [20].
Fruit and vegetable consumption was measured using a self-reported questionnaire. The classification of consumption was based on the 2013 Australian Dietary Guidelines developed by the National Health and Medical Research Council (NHMRC) [21]. The guidelines recommend a minimum daily intake of two servings of fruits and five servings of vegetables, with specific requirements varying by age and sex. Respondents reported their usual daily intake in servings, excluding all drinks, beverages and juices. Individuals aged 15 years and over were also asked about their regular use of tobacco products during the interview. Tobacco products included manufactured (packet) cigarettes, roll-your-own cigarettes, pipes, cigars or other tobacco products, but excluded chewing tobacco and the smoking of non-tobacco products such as marijuana [18]. In this study, a person’s smoking status was categorised as a current daily smoker, defined as someone who reported regularly smoking one or more cigarettes, pipes, cigars or other tobacco products per day.
Excessive alcohol consumption was defined as exceeding the Australian Adult Alcohol Guidelines based on two criteria: component A: consuming more than 10 standard drinks in the past week and component B: consuming more than 4 standard drinks on at least 12 days in the past year [18]. For component A, weekly consumption was estimated using self-reported data on alcohol intake over the 3 most recent drinking days and the total number of drinking days in the prior week. The total alcohol consumed for each drink type was summed, averaged and scaled to reflect weekly consumption. For component B, respondents reported the number of occasions they drank 5 or more standard drinks in a single day over the past year. Excessive drinking was classified as consuming over 21 units (168 g or 213 ml) of alcohol weekly, based on criteria from Lancet Public Health [22]. Additionally, sugary drink consumption included soft drinks (including those in ready-to-drink alcoholic beverages), cordials, sports drinks and energy drinks, while excluding fruit juice, flavoured milk, sugar-free drinks and hot beverages.
Geographic factors included remoteness and area-level socioeconomic disadvantages. Remoteness was grouped as ‘urban cities’, ‘outer regional’, ‘inner regional’, ‘remote’ or ‘very remote’ areas. This classification was based on the Australian Statistical Geography Standard (ASGS). Socioeconomic disadvantage was calculated at the Statistical Area 1 (SA1) and categorised into quintiles from most to least disadvantaged.
Statistical analysis
All data were accessed in the ABS DataLab ( and analysed using the STATA software (version 18, Stata Corp, College Station, TX, USA).
The NATSIHS was designed to produce reliable estimates for the Aboriginal and Torres Strait Islander population, including by state, territory and remoteness area. Each person or household was assigned a weight reflecting the number of people or households represented in the entire population. This weight was based on their probability of selection in the sample. The 2018–2019 dataset included two weight variables: household weight (FINHHWT), used for estimating households, and person weight (FINPERWT), used for estimating the total population [18]. For this study, we used the person-level weight variable (FINGERWT), and the sample size was approximated by dividing the personal weight variable by 100.
Initial analyses involved describing sociodemographic, health risk and geographic factors using frequencies and percentages, all weighted using the personal weight variable. The prevalence of central obesity was calculated across explanatory variables. Subsequently, multilevel multivariable logistic regression models were used to investigate associations between sociodemographic factors, health risk factors and geographic factors with central obesity. Similarly, multilevel regression modelling was used to examine the relationship between normal-weight central obesity and cardiometabolic disorders, such as type 2 diabetes, hypertension, high cholesterol and heart diseases, after adjusting for sociodemographic, health risk and geographic factors.
Multilevel modelling was selected due to its advantages over classical single-level logistic regression models [23, 24]. Specifically, it accounts for the hierarchical nature of the data, recognising that individuals within the same neighbourhood may share similarities in their cardiometabolic risks. Failing to acknowledge these hierarchical structures can lead to an underestimation of standard errors of regression coefficients, potentially inflating the statistical significance of results. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated to measure the association between study factors and outcome variables.
link