ACCR Data Sources and Methdology

Data Sources



Data Sources

Australian Bureau of Statistics (ABS)

The following information was obtained from the ABS:

  • Australian Standard Population (2001) – used in the calculation of age-standardised rates.(1)
  • Estimated resident population – used as the denominator for calculating rates.(2)
  • Population mortality data – used in relative survival calculations.(3, 4)

Australian Childhood Cancer Registry (ACCR)

The ACCR is a complete register of all childhood cancer cases (defined here as children between the ages of 0-14 years) diagnosed in Australia since 1983, except for keratinocyte (non-melanoma) skin cancers. It is funded and managed by Cancer Council Queensland and is supported by the Australasian Association of Cancer Registries. The ACCR is one of the few national population-based registries specifically for childhood cancer in the world.

With appropriate ethical and legislative approvals, every State and Territory Cancer Registry provides information to the ACCR each year on all incident childhood cancer cases. Under these ethics agreements, state-specific data cannot be published without permission.

Patients’ medical charts are also reviewed to collect additional clinical information during on-site visits by the ACCR Data Manager to each of the major children’s hospitals throughout Australia. Mortality status is determined by matching cases in the ACCR against the National Death Index, which contains records of all deaths that have occurred in Australia since 1980.

The information available currently does not include cancer stage (a measure of how far the cancer has spread at the time of diagnosis). However, information to assign stage is now being collected for many types of childhood cancer and it is planned that stage at diagnosis will eventually be available in future releases of the Australian Childhood Cancer Online Statistics.

Note that data for New South Wales and the Australian Capital Territory for 2017 were unavailable at the time of publication and have been approximated based on 2016 data.

Classification of childhood cancers

The tissue of origin of childhood cancers is the best predictor of tumour behaviour and prognosis, and they are classified accordingly. This differs from adult cancers, which are generally classified based on the body site where the cancer occurs. The International Classification of Childhood Cancers, 3rd edition (ICCC-3) (5) is the accepted international standard. All malignant neoplasms are included except for keratinocyte (non-melanoma) skin cancers, which are not registered in Australia and hence are not included in the data contained within the ACCR. In accordance with the ICCC-3, diagnostic groups III (tumours of the central nervous system) and X (germ cell tumours) also include intracranial and intraspinal tumours of benign or uncertain behaviour.



Age-standardised rates

Age-standardised rates adjust for variation in age structures in different populations (either different geographical areas or the same population across time). All incidence and mortality trends shown here were calculated using directly standardised rates. The method involves applying age-specific rates in five-year age groups from the population of interest (i.e. children with cancer in Australia) to a standard population, which on this website is the Australian Standard Population 2001.

Annual percentage change (APC)

This is the annual increase or decrease in the incidence or mortality trends over the specified period. Negative APC values describe a decreasing trend and positive APC values describe an increasing trend. A trend is taken to be statistically significant if the 95% confidence interval does not include zero.

APC values were calculated from a statistical method called joinpoint analysis, using software developed by the Statistical Research and Applications Branch of the National Cancer Institute (see The joinpoint method evaluates changing trends (both the direction and the magnitude of the trend) over successive segments of time. A joinpoint is the data point (year) at which the trend changes significantly.

The analysis begins with the assumption of constant change over time (i.e. no joinpoint). Up to two joinpoints were tested in each model, depending on the number of years of data available and the stability of the yearly estimates. A minimum of 5 years was specified between joinpoints or between a joinpoint and either end of the data series. The selected trend line was the one with the fewest joinpoints which provided the best fit to the observed data, based on Monte Carlo permutation tests.(6)

Australian Standard Population (2001)

The population currently used for direct age-standardisation within Australia is the 2001 Australian Standard Population, which is based on the age distribution of the estimated resident population as at 30 June 2001. The data is available from the Australian Bureau of Statistics.(1)

Confidence intervals

All estimates are calculated with some degree of uncertainty. This uncertainty is typically reported in terms of a confidence interval, which specifies a range of values in which the true data point is expected to occur with a given level of certainty. For example, a 5-year survival rate may be estimated as 85.0% with a 95% confidence interval of 82.5% – 88.1%. This means that there is a 95% probability that the true survival rate will be somewhere between 82.5% and 88.1%.


Incidence measures the number of new cases of cancer diagnosed within a specified population during a given time period (usually one year). Incidence is also commonly expressed as a rate (e.g. per 100,000 population). Since the risk of most cancers varies with age, it is common practice to age-standardise incidence rates to account for variation in the age structures of different populations or changes over time (see Age-standardised rates).


Mortality measures the number of deaths caused by cancer within a specified population during a given time period (usually one year). Similar to incidence, mortality can also be expressed as a rate (per 100,000 population), and these rates are often age-standardised to account for variation in the age structures of different populations or changes over time (see Age-standardised rates).


Survival measures the percentage of people with cancer who remain alive for a given period of time (typically 5 or 10 years). Relative survival is the most commonly presented measure of cancer survival when using data from population-based cancer registries.(7) It compares the observed survival of children with cancer against the expected survival of children from the general population, taking into account age, sex and year of diagnosis. The method does not require knowledge of the specific cause of death, only knowledge of whether the patient has died.

Relative survival estimates can be calculated using either the period or cohort methods. The cohort approach was used for the relative survival estimates shown here. Although the period method is recognised as providing more up-to-date survival estimates,(8) the cohort method was considered more appropriate within the context of children’s cancers to enable the monitoring of changes in treatment over time.

A suite of Stata programs developed by Paul Dickman from the Karolinska Institutet in Sweden were used to generate the relative survival estimates (see These programs use a life table (or actuarial) method for calculating observed survival. This approach involves dividing the total period of observation into a series of discrete time intervals. The survival probabilities are then calculated for each of these intervals, and multiplied together to get the estimate for observed survival. Expected survival (based on total Queensland mortality data obtained from the Australian Bureau of Statistics) was calculated based on the Ederer II method.(9) Three-year averages for expected survival were used to minimise the effects of year to year variation. Relative survival was then obtained from the ratio of observed survival to expected survival. Survival time was censored for patients who were still alive at 31st December 2015.


  1. Australian Bureau of Statistics. Population by age and sex – 2001 census edition. ABS Cat No 3201.0. Canberra: ABS; 2003.
  2. Australian Bureau of Statistics. National, State and Territory Population. Canberra: ABS; 2020. Available at, accessed 7 Jan 2021
  3. Australian Institute of Health and Welfare. General Record of Incidence of Mortality (GRIM) books. Canberra: AIHW; 2020. Available at, accessed 7 Jan 2021.
  4. Australian Bureau of Statistics. Deaths, Australia. Canberra: ABS; 2020. Available at, accessed 7 Jan 2021.
  5. Steliarova-Foucher E, Stiller C, Lacour B, Kaatsch P. International Classification of Childhood Cancer, third edition. Cancer. 2005;103(7):1457-67.
  6. Kim H, Fay M, Feuer E, Midthune D. Permutation tests for joinpoint regression with applications to cancer rates. Statistics in Medicine. 2000;19:335-51.
  7. Dickman PW, Sloggett A, Hills M, Hakulinen T. Regression models for relative survival. Stat Med. 2004;23(1):51-64.
  8. Brenner H, Gefeller O, Hakulinen T. Period analysis for ‘up-to-date’ cancer survival data: theory, empirical evaluation, computation realisation and applications. Eur J Cancer. 2004;40:326-35.
  9. Ederer F, Axtell LM, Cutler SJ. The relative survival rate: a statistical methodology. NCI Monogr. 1961;6:101-21.