|Page tools: Print Page Print All RSS Search this Product|
YEAR OF OCCURRENCE
Year of occurrence versus year of registration
The Australian Bureau of Statistics generally present cause of death data on the basis of registration year. In Australia, most deaths are registered in the calendar year in which they occur, and so share the same year of registration and year of occurrence. However, a small proportion of deaths occur in a given calendar year which are not registered until subsequent years. Approximately 4% to 7% of deaths which occurred from 1997 onwards were registered in subsequent years, and as such will have a year of occurrence that is different to the year of registration.
Year of registration data offers a timely and stable dataset, which will not be altered over time. Conversely, a count of deaths by year of occurrence is more fluid. As deaths are registered and processed, the number of deaths on a year of occurrence basis may continue to rise, and may do so for many years. For example, there are nine currently known deaths which occurred in Australia between 1920 and 1950, which were not registered until 2000 or later. However, despite its fluidity, data presented on a year of occurrence basis has a number of benefits. These include:
Year of occurrence in time series data
Registration lags are more likely among those deaths which are referred to a coroner. These include many deaths due to external causes (such as suicides, transport accidents, and assaults). From 1997 onwards, between 6% and 13% of coroner-certified deaths experienced a registration lag, compared to 4% to 6% of deaths certified by a doctor. As the registration lags have more potential to impact time series for external cause deaths, the potential for year of occurrence data to smooth out these effects is also greater.
Figure 1 shows a time series for suicide deaths on a year of occurrence and year of registration basis. It illustrates the advantage that year of occurrence data has in smoothing the stepwise year on year changes that are noticeable in the year of registration data. In areas that are very closely monitored, such as counts of suicide deaths, this type of analysis can prevent over interpretation of year on year changes that are actually administrative artefacts rather than true changes in counts.
Footnote(s): (a) The data by year of occurrence are based on the year in which the death occurred. A proportion of deaths occur in a given calendar year but are not registered until subsequent years. The year of death data presented in this table may increase in death counts over time, however any changes are likely to be minimal. (b) All causes of death data from the 2006 reference year onward are subject to a revisions process - once data for a reference year are 'final', they are no longer revised. Affected data in this table are: 2006-2013 (final) and 2014 (revised). See Explanatory Notes 52-55, A More Timely Annual Collection: Changes to ABS Processes (Technical Note) and Causes of Death Revisions, 2013 and 2014 (Technical Note) in this publication. (c) Intentional self-harm [suicide], includes ICD-10 codes X60-X84 and Y87.0. Care needs to be taken in interpreting figures relating to suicide. See Explanatory Notes 85-93.
Seasonal patterns in causes of death
Considering deaths by occurrence date also offers a more accurate depiction of seasonal patterns in causes of death. One cause of death known to have a strong seasonal pattern is influenza. In 2014, there were 260 deaths in Australia that were due to influenza. Of these, 93 (35.8%) occurred in the month of August and a further 78 (30%) occurred in September. Figure 2 shows the monthly counts of deaths due to influenza by occurrence and by registration date. This clearly highlights that many of the deaths that occurred in August and September were registered in September and October, a month after the virus was at its most virulent.
During the colder months (May to October) the Department of Health releases Australian Influenza Activity Updates on a fortnightly basis, to monitor potential flu outbreaks. Monitoring of diseases is often focussed on quickly identifying epidemics, so identifying an unusually high prevalence of a disease is reliant on a good understanding of previous seasonal patterns. The timing of flu deaths on a month of death (as opposed to month of registration) basis, provides this more accurate depiction of previous seasonal patterns.
Footnote(s): (a) The data by month of occurrence presented in this table may be affected by lagged registrations. There may be an increase in death counts over time, however any changes are likely to be minimal. (b) All causes of death data from the 2006 reference year onward are subject to a revisions process - once data for a reference year are 'final', they are no longer revised. The month of registration data presented in this table are revised.See Explanatory Notes 52-55, A More Timely Annual Collection: Changes to ABS Processes (Technical Note) and Causes of Death Revisions, 2013 and 2014 (Technical Note) in this publication.
LAGGED REGISTRATIONS: VARIATIONS BY CAUSE OF DEATH AND INDIGENOUS STATUS
In Australia, the number of deaths which experience a registration lag is relatively low (currently 6.7% for deaths which occurred in 2014). It is very low when considering deaths registered two or more years after the occurrence of the death (around 0.1% to 0.2%). However, as previously shown, variations in the magnitude of registration lags can occur for particular causes of death and in certain subsets of the population.
The table below shows the number of deaths by cause that occurred in Australia in 2014, and which were registered in 2014 or 2015. The table indicates a higher proportion of lagged registrations for causes of death relating to the perinatal period (13.9%), causes not elsewhere classified (10.4%), congenital malformations, deformations and chromosomal abnormalities (9.2%) and external causes of death (9.1%). There are differences in the administrative processes that surround these deaths which are likely to influence the timeliness of death registration.
Among deaths of Aboriginal and Torres Strait Islander Australians, 12.2% experienced a lagged registration, compared to 5.4% for the non-Indigenous population. The proportion of deaths with a lagged registration was highest for Aboriginal and Torres Strait Islander persons who resided in very remote areas (21.4% of deaths had a lagged registration, compared with 7.8% for those in major cities). The higher proportion in very remote areas may be influenced by more limited access to services.
NB This data focuses on Aboriginal and Torres Strait Islander deaths for which the usual residence of the deceased was in New South Wales, Queensland, South Australia, Western Australia or the Northern Territory. Data for Victoria, Tasmania and the Australian Capital Territory have been excluded in line with national reporting guidelines (for information on issues with Aboriginal and Torres Strait Islander identification, see Explanatory Notes 56-66).
These documents will be presented in a new window.