Review of 2021 Census Topics

The Australian Institute of Health and Welfare (AIHW) supports the inclusion of a new question on long-term health conditions in the 2021 Census and provides the following comments for consideration.
The term ‘long-term health conditions’ is used throughout this submission, however, there are many synonymous terms such as chronic conditions or non-communicable diseases. Currently defined in the Australian Bureau of Statistics (ABS) National Health Survey (NHS), a long-term health condition is a condition that has lasted, or is likely to last, 6 months or longer.
Long-term health conditions are the leading cause of illness, disability and death in Australia; however, for many long-term health conditions there is no accurate measure of the prevalence of these conditions in the Australian population using existing data sources. In addition, the current data sources are unable to measure the prevalence of many conditions at lower levels of geography such as SA2, SA3, SA4 or Primary Health Network (PHN) and for specific priority population groups, both of which can impact resource allocation, health planning and decision making.
The AIHW acknowledges that work is required to determine the most appropriate and effective way to ask a question on long-term health conditions in the current Census format in order to obtain accurate results. The AIHW would welcome the opportunity to work with the ABS to provide advice on specific details of the question and information items collected.

Information on the prevalence of long-term health conditions is of national importance. Long-term health conditions are the leading cause of ill health, disability and death in Australia and they place a significant burden on the health sector. Examples of long-term health conditions include arthritis, diabetes, asthma, chronic kidney disease, mental health conditions and dementia.
Long-term health conditions have become increasingly prevalent, in part due to the ageing Australian population and social and lifestyle changes. Disease burden from long-term health conditions is significant for the Australian population (accounting for around two-thirds of total burden) (AIHW 2016) and their management is a major challenge within the health system globally (Beaglehole et al. 2011). The importance of this burden is reflected in the development of the National Strategic Framework for Chronic Conditions and various current national and state-based strategies, policies and programs.
The inclusion of a question on long-term health conditions in the Census would enable the prevalence of some long-term health conditions to be calculated and would provide important information that cannot be ascertained from current national health data collections. The information obtained from this question would contribute to health policy development, health program management and could be used for decisions on resource allocation.

References:
AIHW 2016. Australian Burden of Disease Study: Impact and causes of illness and death in Australia 2011—summary report. Australian Burden of Disease Study series no. 4. BOD 5. Canberra: AIHW.
Beaglehole R, Bonita R, Horton R, Adams C, Alleyne G, Asaria P et al. Priority actions for the non-communicable disease crisis. The Lancet 377:1438-47.

An accurate measure of the prevalence of some long-term health conditions in the Australian population is not available, for example, for chronic obstructive pulmonary disease and dementia. The prevalence of specific individual diseases at a national level is often low and existing survey-based data sources, such as the ABS National Health Survey, are unable to provide estimates of the prevalence of these rarer conditions. In addition, population groups including Indigenous Australians, people residing in very remote locations and institutionalised individuals are often excluded from or underrepresented in these surveys. The results, therefore, often have high margins of error and there is little confidence in the accuracy and representativeness of the estimates.
Work undertaken by AIHW and others also show that the prevalence of long-term health conditions vary markedly by geography. Health planning is now largely undertaken at the small area level (for example by Primary Health Networks) and data on health need by this level of disaggregation or lower is required. Currently, the prevalence of many long-term health conditions can only be reported at the national or state and territory level. This potentially masks differences that could exist across the country given the diverse nature of Australia in terms of remoteness and socioeconomic position, both of which have been shown to strongly influence the health need of population groups.
The ability to combine the information collected from a question on long-term health conditions with the other characteristics collected in the Census (such as education, Indigenous status and country of origin) would enhance the information and enable analysis that would strongly assist policy development and planning purposes. For example, for some long-term health conditions, including dementia and other neurological conditions, chronic obstructive pulmonary disease, some cardiovascular diseases (in particular atrial fibrillation and flutter) and Type 1 diabetes, the prevalence in the Australian Indigenous population is unknown. Including a health condition question on the Census would help to fill this important information gap. Collecting information on health conditions in the Census may also help improve the accuracy of modelled small area estimates for other collections such as ABS population surveys.

The topic needs thorough consideration and testing to ensure the question is asked in a way that is easily understood by the respondent and that the current Census format (i.e. a self-reported, or reported on behalf of other household members, household survey) would support collection of the information obtained. There are questions about the validity of self-reported data for some diseases, and this needs to be examined further. Further analysis of measured and self-reported data from the 2011–13 Australian Health Survey could be useful for this, along with a review of literature on the validity of self-reported data for particular conditions.
A specific example of long-term health conditions that are not recommended for inclusion are mental health conditions. Including a mental health question is not considered to be particularly useful as the results would be difficult to interpret i.e. depending on household composition, a third person would be completing information on the rest of the members of the household. There has been previous work done that identified a disconnect between ABS self-report surveys which ask questions about mental health and the gold standard prevalence measures (The Australian National Survey of Mental Health and Wellbeing [NSMHWB]) which use a diagnostic questionnaire like the World Health Organization World Mental Health Composite International Diagnostic Interview (WHO WMH-CIDI). Self-reported measures have been observed to produce widely different mental health prevalence rates depending on the survey sample and question item stem.
With that said, there are examples of other health surveys that have used a paper-based survey that the respondent filled out themselves (for example, the AIHW’s Coordination of health care study) and the ABS have significant expertise in this area to ensure a robust question is included and robust data items are obtained. Work on question design would not need to occur from scratch and there are many examples that could be drawn upon. In fact, it would be important to align the question and responses with other existing health surveys so that information could be compared and used to validate the information collected in the smaller, more targeted, health surveys.
If necessary, AIHW would be willing to provide advice to the ABS in the design of the question to help ensure its accuracy and usefulness.

The topic of long-term health conditions is collected in other formats and has not been shown to be considered intrusive or offensive to respondents. It will be important to consider the type of long-term health conditions included in the question as some may be considered more intrusive than others (for example, mental health conditions and some disabilities) and may not be appropriate to ask on behalf of other members in a household because they may not be aware of each other’s health conditions. This issue is both in terms of maintaining the right to privacy of individuals as well as the ability of one member of the household to accurately report information on behalf of all members. The Census already allows for respondents to request separate forms/IDs for privacy, so there may already be solutions in place to mitigate this issue.

A question on long-term health conditions needs to be simple and the prior work undertaken to design a question for other surveys could be utilised. Some level of instruction and explanation of terms would be required but these should not need to be lengthy as terminology should be understood particularly by people with the condition. If necessary, the number of response categories could be restricted to only those long-term health conditions deemed the highest priority for which the Census collection methodology is considered suitable.
Any proposed long-term health conditions in the Census would be codified using classifications based on the International Classification of Diseases: 10th Revision (ICD-10). Where possible, data on long-term health conditions should be collected to be comparable to the National Health Survey (NHS). The code list used for the 2014–15 NHS was initially compiled for use in the 2001 NHS by the Family Medicine Research Centre, University of Sydney, in association with the ABS. This was also used for the 2004–05, 2007–08 and 2011–12 NHS. Conditions classified at the full level of detail are not generally available for output from the survey; however, they can be regrouped in various ways for output.
As an example, the question format might be worded as:
Have you ever been told by a doctor that you have or have had?
Yes, Arthritis
Yes, Osteoporosis
Yes, Asthma
Yes, Chronic Obstructive Pulmonary Disease/Emphysema
Yes, Cancer
Yes, Diabetes (apart from during pregnancy)
Yes, Heart attack or angina
Yes, Stroke
Yes, Chronic kidney disease
Yes, Mental health condition, including depression or anxiety
Yes, Dementia
Yes, Long-term injury
Yes, Other long-term condition

The need for information on the prevalence of long-term health conditions is likely to continue to increase as the population ages. The need to understand the prevalence of long-term health conditions at small geographic levels and be able to plan interventions is not likely to diminish in the near future.
Information collected at 5-yearly intervals will provide important time-series information allowing positive or negative changes to be measured and the effect of population-based interventions to be assessed. The information collected via the Census will also underpin and support information from other sources.

There has been significant national investment in the collection of health-related information for the purposes of research, policy development and resource planning. Data sources including surveys, jurisdictional and national disease registers and administrative data collections provide crucial insights and information about the health status of individuals and the services they receive. However, the need for more information and increased investment in health information infrastructure remains high.
Soon, Australia will have (for the first time) nation-wide integrated health information from a number of administrative data sources including hospital admitted patient care services, emergency department and outpatient services, Pharmaceutical Benefits Scheme and Medicare Benefits Schedule which will allow for powerful investigation of the ‘serviced’ or ‘treated’ population. However, the question of health need (health status of individuals who are receiving treatment and those who are not) will remain unanswered unless included in national collections.
Given that there are existing survey-based data sources collecting information about some long-term health conditions, the range of conditions to be included in the question needs to be carefully considered and perhaps restricted to those where significant information gaps exist and/or conditions that can be reliably collected via the Census.