Notes on the data: Self-assessed health

Estimated population, aged 15 years and over, with fair or poor self-assessed health, 2017-18


Policy context:  Self-assessed health status is commonly used as a proxy measure of actual health status; and how people rate their health is strongly related to their experience of illness and disability [1,2]. This measure is therefore an important indicator of key aspects of quality of life [3].

Australians generally consider themselves to be healthy. In 2017–18, over half (56.4%) of Australians aged 15 years and over rated their health as ‘very good’ or ‘excellent’, while only 3.7% rated it as ‘poor’; a further 11.9% rated it as 'fair' [4]. Older Australians generally rated themselves as having poorer health than younger people, with persons aged 75-84 years and 85 years and over recording the highest proportions of fair or poor health, at 30.9% and 35.8% respectively [4]. There was little difference in the way men and women assessed their overall health, with men slightly more likely to report their health as fair or poor other than at ages 85 years and over, where almost one third more women than men reported poorer health[4].


  1. Australian Bureau of Statistics (ABS). Profiles of health, Australia, 2011-13. (ABS Cat. no. 4338.0). Canberra: ABS; 2013.
  2. Doiron D, Fiebig DG, Johar M, Suziedelyte A. Does self-assessed health measure health? Sydney, NSW: UTS; 2014.
  3. McCallum J, Shadbolt B, Wang D. Self-rated health and survival: a seven-year follow-up study of Australian elderly. Am J Public Health. 1994;84(7):1100-5.
  4. Australian Bureau of Statistics (ABS). Self-assessed health. National Health Survey: First Results, 2017-18. (ABS Cat. no. 4364.0.55.00). Canberra: ABS, 2018. Available from:; last accessed 17 December 2019


Differences from data published by the Australian Bureau of Statistics (ABS):

Data by quintile of socioeconomic disadvantage and Remoteness will differ to the extent that data extracted from Survey TableBuilder have been randomised, whereas those published by the ABS are not. In addition, rates published by the ABS for modelled estimates are generally crude rates; rates published by PHIDU are age-standardised.

Modelled Estimates:

In the absence of data from administrative data sets, estimates are provided at the Population Health Area (PHA) level for selected risk factors from the 2017–18 National Health Survey (NHS), conducted by the Australian Bureau of Statistics (ABS). The estimates at the Population Health Area (PHA) level are modelled estimates produced by the ABS (estimates at the Local Government Area (LGA) and Primary Health Network (PHN) level were derived from the PHA estimates).

Details of the method used and accuracy of results are available from the ABS paper Explanatory Notes: Modelled estimates for small areas based on the 2017-18 National Health Survey, available here.

For the Primary Health Network (PHN) data, differences between the PHN totals and the sum of LGAs within PHNs result from the use of different geographic correspondence files.

Estimates for quintile of socioeconomic disadvantage of area and Remoteness Area are direct estimates, extracted using the ABS Survey TableBuilder.

Users of these modelled estimates should note that they do not represent data collected in administrative or other data sets. As such, they should be used with caution, and treated as indicative of the likely social dimensions present in an area with these demographic and socioeconomic characteristics.

The numbers are estimates for an area, not measured events as are, for example, death statistics. As such, they should be viewed as a tool that, when used in conjunction with local area knowledge and taking into consideration the prediction reliability, can provide useful information that can assist with decision making for small geographic regions. Of particular note is that the true value of the published estimates is also likely to vary within a range of values as shown by the upper and lower limits published in the data (xlsx) and viewable in the column chart in the single map atlases.

What the modelled estimates do achieve, however, is to summarise the various demographic, socioeconomic and administrative information available for an area in a way that indicates the expected level of each health indicator for an area with those characteristics. In the absence of accurate, localised information about the health indicator, such predictions can usefully contribute to policy and program development, service planning and other decision-making processes that require an indication of the geographic distribution of the health indicator.

The relatively high survey response rate in the NHS provides a high level of coverage across the population; however, the response rate among some groups is lower than among other groups, e.g., those living in the most disadvantaged areas have a lower response rate than those living in less disadvantaged areas. Although the sample includes the majority of people living in households in private dwellings, it excludes those living in the very remote areas of Australia and discrete Aboriginal and Torres Strait Islander communities; whereas these areas comprise less than 3% of the total population, Aboriginal people comprise up to one third of the population. The survey does not include persons usually resident in non-private dwellings (hospitals, gaols, nursing homes - and also excludes members of the armed forces).

This and other limitations of the method mean that estimates have not been published for PHAs with populations under 1,000.

The ABS used a number of methods to measure the quality of the estimates, one of which is the relative root mean squared error (RRMSE) of the modelled estimates. The RRMSEs are included with the data. Users are advised that:

  • estimates with RRMSEs less than 25% are considered reliable for most purposes;
  • estimates with RRMSEs from 0.25 and to 0.50 have been marked (~) to indicate that they should be used with caution; and
  • those greater than 0.50 but less than 1 are marked (~~) to indicate that the estimate is considered too unreliable for general use.

Indicator detail

The data on which the estimates are based are self-reported data, reported to interviewers in the National Health Survey. Respondents aged 15 years and over were asked to assess their health on a scale from 'poor' to 'excellent' (the scale was 'poor', 'fair', 'good', 'very good', or 'excellent'). The data reported are the sum of responses categorised as 'poor' or 'fair'.


Geography: Data available by Population Health Area, Local Government Area, Primary Health Network, Quintile of socioeconomic disadvantage of area and Remoteness Area


Numerator:  Estimated number of people aged 15 years and over reporting their health as 'fair' or 'poor' (and not 'good', 'very good' or 'excellent')


Denominator:  Population aged 15 years and over


Detail of analysis:  Indirectly age-standardised rate per 100 population (aged 15 years and over); and/or indirectly age-standardised ratio, based on the Australian standard



PHA, LGA & PHN: Age-standardised rates are based on Australian Bureau of Statistics data, produced as a consultancy for PHIDU, from the 2017—18 National Health Survey.

Quintiles & Remoteness: Compiled by PHIDU based on direct estimates from the 2017-18 National Health Survey, ABS Survey TableBuilder.


© PHIDU, Torrens University Australia This content is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Australia licence.