Statistical methods
Count
Count refers to the number of events (e.g., persons with disease, hospitalisations, deaths) occurring in the population or the number of the people residing in a specific geographic area. Please note that for hospitalisation and deaths data, counts were combined over a multiple-year period, due to the small number of counts at the SA2 level. Mean number of events per year was calculated by dividing the combined counts by the number of years included in the time period. Note that some disease and risk factor prevalence counts came from direct and modelled survey estimates. These counts include a level of error related to the sample size of the survey. For more details of the method to derive error and their related confidence, consult the following documents ABS NHS 2017–18 Modelled estimates for small areas: Explanatory notes in Technical notes.
Crude rate
A crude rate is the number of events in a given period divided by the size of the population at risk in the specific time period and approximated by the ERP for death and hospitalisation rates and the weighted survey population for prevalence. This period can be a year or multiple years. For example, CKD death rates were calculated based on the combined deaths and populations over the 5-year period.
Age-specific rate
An age-specific rate is a rate for a specific age group, where the numerator and denominator relate to the same age group. It measures the occurrence of an event within a specific age range in the population. In this product, 3 broad age groups were used for reporting: 0–54 or 18–54; 55–74; and 75 and over (survey prevalence estimates only).
Age-standardised rates
Age standardisation is a method of reducing the influence of age when comparing populations with different age structures. Directly age-standardised rates were computed as shown below to enable the comparison of rates between geographical areas.
Direct method
A directly age-standardised rate is derived by applying the age-specific rates in the study population to a single standard population—the June 2001 Australian ERP. Rates that have been directly age-standardised are comparable across geographic areas. Direct age-standardisation was applied to prevalence estimates at the state/territory level, and for all hospitalisation and deaths data.
The computational formula used to derive the directly age-standardised rate (ASR) is:
with:
- Ni the standard population for each age group i
- pi the event rates for each age group i.
95% confidence interval
The computational formulae used to derive the standard error (SE) and 95% confidence interval (CI) of the ASR are:
Note: Assumes Poisson distribution of the events and normal distribution of the ASRs.
Rate ratio
The rate ratio (RR) provides a measure of the relative gap in rates (crude and age-standardised) between 2 populations. Directly age-standardised RRs are the rates in the geographical areas (state/territory, PHN, PHA) divided by the national ASR.
The computational formula used to derive the RR between 2 ASRs (e.g., ASRa—local rate and ASRb—national rate) is:
95% confidence interval
In order to assess whether the relative difference between the 2 rates is a true difference, the 95% CI for the RR (i.e., CI(RR)) was calculated using the formula below (Rothman et al. 2008):
Where:
With:
- e as exponential
- SE as standard error
- ln as natural logarithm.
Note: Assumes Poisson distribution of the events and normal distribution of the ASRs.
Percentile
Percentile is a measure derived by ranking the geographic areas (PHA) by the ASR and dividing it into 100 (percentile) equal parts. Percentile can also mean the cut-off points that make these divisions. In this report, percentile rankings are allocated for PHA hospitalisation and deaths data—the higher the ranking group number the higher the rate.
ABS surveys are designed to obtain a sample that represents the characteristics of the national population on a smaller scale, and thus to reduce the cost and burden associated with data collection.
Direct survey estimates
Estimates for states and territories and the PHNs encompassing a whole state or territory—Tasmania, the Northern Territory and the Australian Capital Territory—were based on direct estimates. Modelled estimates were used for Tasmania, the Northern Territory and the Australian Capital Territory where their direct age-specific estimates were found unreliable.
The sampling designs for the 2011–12 AHS and the 2017–18 NHS were not intended to provide estimates by small areas (PHN and PHA). To overcome this limitation, the ABS undertook modelling to produce synthetic estimates for smaller geographical areas.
Modelled survey estimates
The modelled estimates were based on random effects logistic regression models fitted to data from the 2017–18 NHS, ERP as at 30 December 2017, 2016 Census of Population and Housing and administrative data, adjusted (where possible) to match the scope of the NHS 2017–18.
For both surveys (AHS and NHS), areas were excluded from the modelling process if:
- they were classified as Very remote
- they were classified as discrete Aboriginal and Torres Strait Islander communities
- they had zero residents.
Data for the Northern Territory, in particular, should be interpreted with caution as Aboriginal and Torres Strait Islander communities comprise around 28% of the resident population of the Northern Territory of which almost 8 in 10 Indigenous Australians live in Remote (21%) or Very remote (56%) areas (ABS 2017).
For detailed information about modelled estimates and the methodology applied, please refer to the following papers in Technical notes:
- National Health Survey, 2017–18 modelled estimates for small areas – explanatory notes
- Australian Health Survey 2011–12 modelled estimates for small areas – explanatory notes