Australian Bureau of Statistics (ABS) mortality data are the main source of suicide statistics in Australia. The ABS is part of a complex process that generates these statistics. The two other main parts of the system are coroners and the National Coroners Information System (NCIS). Coding rules that form part of the International Classification of Diseases (ICD), which the ABS is required to apply, also affect the statistics.

In recent years, the ABS has cautioned that suicide data may be underestimated and ‘observed changes over time are likely to have been affected by delays in [coroners] finalising a cause.’

Investigation of recent suicide data

This report investigated deaths occurring in 2004 mainly using cases extracted from the NCIS as at early 2008. Comparisons were made using ABS data added to the NCIS, including Underlying Cause of Death codes, and with aggregate data from the ABS mortality data files.

The comparisons indicated that, at the time that ABS coders assigned ICD-10 codes to deaths, the lack of information from coroners concerning intent impeded their assignment of codes such as those for ISH. The main effect of this is that some deaths that were later finalised by coroners and recorded in the NCIS as being due to ISH were not classified as ISH in the ABS mortality data. Aside from thee effects of incomplete information, the quality of ABS coding appears to have been good.

Taking the misclassified cases into account results in larger numbers of deaths due to ISH than the numbers published in ABS mortality data for the same period. Revised estimates are 3% higher than the ABS value when using similar criteria to the ABS, 11% higher when using these criteria and adding cases closed between the publication of Causes of death, Australia, 2004 (ABS 2006a) in March 2006 and the extraction date of the data used in this project, and 16% higher when including all deaths in NCIS identified as ISH when using a standard definition applied early in 2008.

Future statistics on suicide

The ABS has introduced changes to its cause of death processes, which can be expected to improve the completeness of ISH data for deaths registered from 1 January 2007. The most important change is the introduction of a revision process for causes of death. Data on deaths registered before 2007 were finalised by the ABS before the annual release of the report of the causes of deaths (e.g. by March 2008 for deaths registered in 2006). Under the new system, causes of death can be reviewed for at least 2 years after this (i.e. to early 2011 for deaths registered in 2007), allowing longer for the ABS to receive final information from coroners.

Although it is not possible to provide an exact prediction of the effect that this change will have on the completeness of the data, this study suggests that it is likely that more than half of ISH deaths previously misclassified will be correctly classified in the future.

The inclusion of NCIS deaths meeting certain criteria as being due to ISH (even though a coroner’s finding is not available) and statistical modelling could improve estimates for deaths registered before 2007. For deaths registered after that, methods to improve the timeliness of data and the value of a standard definition of ISH for statistical purposes should be explored with coroners and other stakeholders.