External Causes of Death Data Analysis

The earlier section Potential Data Quality Issues Affecting External Causes of Death Data discussed the effect of incomplete source information on the quality of causes of death codes. This section considers the interpretation of the resulting data, given these issues. The data in this section refer to all persons.

Any increase in the number of cases with incomplete information about causes of death at the point of coding will adversely affect data quality relative to previous years. It is clear that the proportion of coronial cases which are not closed when coding is finalised has been increasing in recent years. In particular, cases which in previous years might have been coded as assault or suicide, might now be more likely to be classified as an accident. While some of the numbers may be relatively small in absolute terms, it is essential to monitor this because of the potential impact on data accuracy.

It is important to note that for the last two reference years, a range of actions were taken to improve the completeness of coronial information at the point of coding in order to ensure that data quality remains at acceptable levels. An indication of how some of these actions improved the quality of the data is available from the 2004 data where we can compare data compiled before and after extra processing time. Some cases were able to be coded more accurately (those with more information available), with higher numbers able to be coded to categories such as suicides (footnote 6).

In broad terms, an increase over time in the number of accidents and a decrease in the number of suicides and assaults may result from a higher proportion of incomplete records. It is difficult, though, to completely rule out whether such increases reflect real world events. There are many different factors which may influence increases or decreases in the number of deaths in any particular category over time, including improvements in processes.

Firstly it is important to note that there is no evidence that there has been any impact on the chapter as a whole. Over the years 2001 to 2005, the proportion of all deaths resulting from external causes has remained at around 6% (6.1% for deaths registered in 2005). The number of deaths from external causes has varied over this time, increasing slightly overall.

There is however some evidence that despite the additional quality assurance measures implemented before the statistics were compiled, an increased number of records with incomplete information for coding has had some impact on the specificity of the coding of some external causes of death cases. The table below indicates the direction of possible changes within external causes of death data over the past five years, which might be related to a lack of information available for coding. These changes are discussed in more detail in the following analysis. Note that it is difficult to quantify the size of any such impact, let alone clearly attribute any change to this factor alone. Future work will continue with the aim of understanding, managing and reducing it where possible.

ACCIDENTS

Over the last five years, the number of deaths resulting from accidents has increased from around 4,800 in 2001 to about 5,300 in 2005. Some accidental causes have increased in number (e.g. falls, exposure to inanimate mechanical forces, exposure to smoke, fire and flames, accidental poisoning by and exposure to noxious substances) while others have decreased (e.g. transport accidents, accidental drowning and submersion) over this period.

	2001 no.	2002 no.	2003 no.	2004 no.	2005 no.
Transport accidents (VO1-V99)	2 004	1 907	1 811	1 689	1 638
Falls (W00-W19)	634	629	709	873	996
Exposure to inanimate mechanical forces (W20-W49)	118	135	115	151	137
Exposure to animate mechanical forces (W50-W64)	13	10	5	8	14
Accidental drowning and submersion (W65-W74)	261	232	201	197	192
Other accidental threats to breathing (W75-W84)	217	222	275	342	260
Exposure to electric current, radiation and extreme ambient air temperature and pressure (W85-W99)	34	27	25	24	29
Exposure to smoke, fire and flames (X00-X09)	66	103	98	86	109
Contact with heat and hot substances (X10-X19)	7	12	12	17	4
Contact with venomous animals and plants (X20-X29)	13	3	5	3	np
Exposure to forces of nature (X30-X39)	23	23	28	65	40
Accidental poisoning by and exposure to noxious substances (X40-X49)	642	568	629	737	740
Overexertion, travel and privation (X50-X58)	6	3	3	np	3
Accidental exposure to other and unspecified factors (X58-X59)	802	1 032	949	1 015	1 104
Total	4 840	4 906	4 865	5 209	5 267
(a) Year of registration
Source: Causes of Death, 2005 (ABS cat.no. 3303.0)

Transport accidents (V01-V99)

Transport accidents are the largest cause of accidental deaths in Australia, representing 31.1% of all accidental deaths registered in 2005. The number of these deaths has decreased since 2001. A decrease over this period is apparent for all of the major categories, although in 2005 there were increases recorded for two of the highest level groups of codes, car occupant injured in transport accident (V40-V49) and motorcycle rider injured in transport accident (V20-V29).

A subset of this data (motor vehicle traffic accidents) can be compared with data from the Australian Transport Safety Bureau (ATSB 2006) on deaths of road users. ATSB statistics are based on reports of road crashes from police and other organisations responsible for road safety within each jurisdiction (footnote 7).

For most of the period 1997 to 2003, data from these two sources follow a very similar pattern, with the ABS causes of deaths data slightly lower than the ATSB data. The year to year variation in the number of deaths is largely consistent in both data collections. However in 2004 and 2005 the causes of death data are not consistent with ATSB deaths data, and are much lower. The difference between the two sources of data is particularly marked for New South Wales.



The decrease in deaths due to transport accidents may be related to a lack of clear information about the mechanism at the time of coding, resulting for example in some cases being allocated to X59 Exposure to unspecified factor.


Falls (W00-W19)

The number of registered deaths attributed to accidental falls has been steadily increasing over the last five years, from 634 in 2001 to 996 in 2005. As a proportion of all deaths recorded as being due to accidents, falls has increased from 13.1% in 2001 to 18.9% in 2005.

This increase occurred mainly in Victoria (rising from 95 cases in 2001 to 431 in 2005), reflecting the introduction of a campaign to minimise the risk of falls and fall-related injuries in that state (which included the referral of more deaths involving a fall to coroners, particularly among older people). The increase was mainly in age groups over 75 years, about two-thirds of the increase was in females, and almost all of the increase was in deaths from accidental falls certified by a coroner. This increase can be expected to continue into the future if other jurisdictions follow suit with similar campaigns.

Accident codes likely to be associated with suicides or assaults

Threats to breathing (W75-W84)

The number of deaths classified to other accidental threats to breathing has varied over the past five years. Overall there has been an increase in this period, with the highest numbers recorded in 2003 and 2004 (followed by 2005).

These codes are of particular interest as there may be a relationship to suicide given that hanging (including strangulation and suffocation) is the most common method of suicide, used in 51% of all suicides registered in 2005. The code W76 Other accidental hanging and strangulation may be related to possible suicides for which a determination of intent is not available for coding purposes, although W75 Accidental suffocation and strangulation in bed, W83 Other specified threats, and W84 Unspecified threat could also be the outcome for such cases. A small number of assaults could also be coded to these categories if information on intent is lacking. There was an increase of around 70 deaths in these particular accident categories between 2001 to 2004, with a return to 2001 levels in 2005 (102). Most of the increase in the number of cases coded to these categories occurred in New South Wales (31 deaths in 2001 to 77 deaths in 2004 and 49 deaths in 2005) and Queensland (30 deaths in 2001 to 69 deaths in 2004 and 23 deaths in 2005), across all adult age groups.

Poisoning (X40-X49)

The number of deaths classified as accidental poisoning by and exposure to noxious substances has also increased over the last five years, although there has been variation over the period. The highest numbers were recorded for the 2004 and 2005 registration years (737 and 740 deaths respectively).

Poisoning (including drugs, chemicals and gases) is also a common method of suicide (representing 27.8% of suicides in 2005), and is also involved in a small number of assault cases. There was a corresponding decrease in suicides associated with this mechanism over the same period (see discussion below). The increase in accidental poisoning deaths between 2001 and 2005 has occurred mainly in age groups between 35 and 69 years, while the decrease in intentional (suicide) poisoning deaths over the same period has occurred mainly in age groups between 20 and 54 years of age. In the 15-24 years age group, the decrease in intentional poisoning deaths has occurred alongside a decrease in accidental poisoning deaths.

Inanimate mechanical forces (W20-W49)

The number of accidental deaths classified as caused by inanimate mechanical forces (e.g. falling objects, sharp glass, knife, machinery, firearms) is relatively small but has increased over the last five years, although the number recorded in 2005 (137) was lower than that recorded for 2004 (151).

Firearm discharge can also be involved in suicides (7% of all suicides in 2005) or assaults (10% of all assaults). Although the number of deaths coded to accidental firearm discharge is relatively small, the number recorded in the 2003, 2004 and 2005 registration years is higher than in 2001 and 2002. This group of codes includes W31 Contact with other and unspecified machinery, W32 Handgun discharge, W33 Rifle, shotgun and larger firearm discharge, W34 Discharge from other and unspecified firearms, and W40 Explosion of other materials. For firearm deaths there has been a decrease in intentional (suicide) firearm deaths and an increase in accidental (unintentional) firearm deaths, noticeable in the last several years. The largest increase in accidental firearm deaths over the last five years was in New South Wales (increase of 19 deaths), while suicide firearm deaths in New South Wales decreased by 58 deaths (around half of the national decrease).

Exposure to unspecified factor (X59)

The number of accidental deaths classified as X59 Exposure to unspecified factor has increased from 802 in 2001 to 1,102 in 2005. These unspecified accidents comprised 14% of all external causes of death in 2005 (up from 10% in 2001).

Almost all of the increase of around 290 deaths classified to X59 in the last five years has been in coroner-certified deaths, with most registered in New South Wales. Most of these deaths occurred in older age groups, but there have been increases in recent years for younger age groups as well.

According to ICD-10 rules, this code is used when information on both the intent and the mechanism of death is missing (but the death is clearly due to an external cause based on the existing information). A lack of information about intent means that these cases will be treated as accidents, but due to the lack of information about the mechanism, they cannot be coded as any particular type of accident. The increase in deaths coded to X59 over the last few years may be related to decreases recorded in suicides, assaults and transport accidents, and may reflect an increasing lack of information about both intent and mechanism of death.


SUICIDES

The codes X60-X84 Intentional self-harm (suicide) include purposely self-inflicted poisoning or injury. The number of cases coded to intentional self-harm has decreased over the last five years, with the number recorded for 2005 (2,101) very similar to that recorded for 2004 (2,098). The number of suicides decreased overall by about 14% between 2001 and 2005. The decrease is most apparent in the 25-29 years, 30-34 years and 35-39 years age groups (decreases of about 29%, 22% and 29% respectively over the period), the 15-19 years age group (decrease of 24%), and the 70-74 years age group (20%).

Decreases in the number of suicides over the past five years occurred in most jurisdictions except for South Australia, where the number of suicides has actually varied over this period. (In 2001 there were 206 deaths from suicide in South Australia, decreasing to 169 in 2002 and increasing overall to 233 in 2005).

For the most common method of suicide (hanging), the number of deaths recorded over the past five years has remained at around 1,000. Deaths allocated to intentional poisoning, firearm discharge or other methods have all decreased over the same period.

	2001 no.	2002 no.	2003 no.	2004 no.	2005 no.
Poisoning by drugs (X60-X64)	285	287	278	229	247
Poisoning by other gases, vapours, chemicals and other noxious substances (incl. motor vehicle exhaust) (X65-X69)	513	439	415	402	337
Hanging, strangulation and suffocation (X70)	1 050	1 045	996	998	1 068
Firearms and explosives (X72-X75)	262	218	194	169	147
Other (b) (X71, X76-X84)	344	331	330	300	302
Total	2 454	2 320	2 213	2 098	2 101
(a) Year of registration. (b) Includes drowning, smoke/fires/flares, sharp objects, jumping from high place, jumping or lying before moving object, other and unspecified means. Source: Suicides, Australia, 2005 (ABS cat. no. 3309.0)

As discussed above under accidents, some possible suicide cases may have been coded as accidents due to mechanisms such as hanging or poisoning if there was no information available to confirm the intent of the death at the time of coding. Of the accident codes listed below as most likely to be used in these instances, the key categories of interest are hanging/strangulation/suffocation, poisoning and firearm discharge.

Description	ICD-10 code
Accidental falls	W13, W15-W17, W19
Contact with knife, sword or dagger	W26
Contact with non powered hand tool	W27
Contact with other powered hand tools and household machinery	W29
Firearm discharge, explosion of other material, contact with other and unspecified machinery	W31-W34, W40
Accidental drownings and submersion	W65, W67, W69, W73-W74
Accidental strangulation/hanging/suffocation, other specified and unspecified threats to breathing	W75-W76, W83-W84
Exposure to unspecified electric current	W87
Exposure to unspecified smoke, fire and flames	X09
Accidental poisoning (incl. drugs, chemicals and gases)	X40-X49
Lack of food	X53
Exposure to unspecified factor	X59

Some statistics on suicides are publicly available from other organisations. Many of these are also based on actual coronial records and report higher numbers of suicide deaths. There are a number of reasons why these data differ from the official statistics published by the ABS. Higher numbers of suicides will result from the use of less strict criteria to classify suicides. Relaxing the burden of proof required to conclusively code a case as a suicide, making assumptions about intent, and not requiring a legal/medical finding, make it easier to code deaths as suicide outside the ABS. Higher numbers of suicide cases will also result from accessing coronial records later in time, thereby allowing more time for coronial cases to be closed and documented.


ASSAULTS

The number of assaults (homicides) recorded relies on coroners making a finding that injuries were inflicted by another person with intent to injure or kill, by any means. As a proportion of all external causes, assaults represented 2.5% of deaths registered in 2005. The most common type of assault causing death in 2005 was assault by a sharp object (X99) with 80 deaths, followed by assault by bodily force (Y04) with 27, assault by blunt object (Y00) with 26, and assault by hanging, strangulation and suffocation (X91) with 24.

Selected codes	2001 no.	2002 no.	2003 no.	2004 no.	2005 no.
Assault by drugs, medicaments and biological substances (X85)	4	4	3	0	np
Assault by gases and vapours (X88)	np	5	5	np	0
Assault by hanging, strangulation and suffocation (X91)	27	19	16	16	24
Assault by drowning and submersion (X92)	5	3	3	np	np
Assault by handgun discharge (X93)	13	12	6	np	4
Assault by rifle, shotgun and larger firearm discharge (X94)	10	20	24	5	10
Assault by other and unspecified firearm discharge (X95)	24	13	24	10	5
Assault by explosive material (X96)	np	9	np	0	0
Assault by smoke, fire and flames (X97)	5	7	9	np	np
Assault by sharp object (X99)	110	98	97	72	80
Assault by blunt object (Y00)	34	40	20	7	26
Assault by crashing of motor vehicle (Y03)	np	3	5	0	7
Assault by bodily force (Y04)	36	31	33	22	27
Assault by other specified means (Y08)	4	4	np	0	np
Assault by unspecified means (Y09)	18	20	29	22	6
Total	300	291	278	164	199
(a) Year of registration. Source: Causes of Death, 2005, (ABS cat. no. 3303.0)

The number of deaths due to assaults has greatly decreased over the last five years, from 300 deaths to around 200. The lowest numbers were recorded in the 2004 and 2005 registration years (164 and 199 deaths respectively, down from 278 in 2003). The decrease between 2001 and 2005 is most evident for New South Wales (109 to 63 deaths), Queensland (60 to 29 deaths) and the Northern Territory (22 to 7 deaths). It should be noted that for the 2005 registration year, the total number of open coroner's cases was relatively high in New South Wales and Queensland.

As discussed above under accidents, some possible assault cases may have been coded as accidents due to inanimate mechanical forces or hanging/strangulation/suffocation if there was no information available to confirm the intent of the death at the time of coding.

These data can be compared with data on the number of victims of recorded crime for the category of murder. Data on victims of recorded crime is based on administrative data relating to murders reported to or detected by police and is published in Recorded Crime - Victims, Australia (cat. no. 4510.0 (ABS 2006c). While the collection methods are entirely different, the resulting figures are roughly similar up to 2003, but for 2004 and 2005 the causes of death data are lower than the recorded crime data.