2. Methods

2.1 Coding Of Injury

In Western Australia, coding of injury is based on the World Health Organisation’s (WHO) Ninth Revision of the Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death (commonly referred to as ICD-9). In July 1988, a clinical modification of ICD-9 (ie. ICD-9-CM) was introduced as the basis for hospital admission diagnostic coding.

In this report, the supplementary classification of external causes of injury and poisoning (E codes) was used to identify bicycle and vehicle related crashes, as well as all injuries. (See appendix table A20).

2.2 Data Sources

2.2.1 Hospitalisations
Hospitalisation data were obtained from the Western Australian Hospital Morbidity Data System (HMDS).

Most injury cases in this database have an ‘Injury and Poisoning’ (Chapter 17) code as principal diagnosis and require the specification of an external cause (E code). For some injury cases, an E code has been specified but the major diagnosis is not found in Chapter 17. In this report, the latter cases have been eliminated from the analysis. That is, only cases with a principal diagnosis within Chapter 17 have been included.

The following procedures were undertaken to prepare the data for analysis:

• Hospital admission records were matched by name, date of birth, sex and postcode to identify episodes of care for the same crash.

• All cases re-admitted within 28 days of a previous admission were eliminated. Most transfers would be removed by this process as, in most cases, they would be considered a re-admission.

• Cases recorded as transfers but with no record of admission at another hospital were retained.

The above process was necessary to ensure that patients were counted once and only once for every injury event resulting in hospitalisation. Re-admission and transfer records were included for hospital length of stay calculations, however.

2.2.2 Deaths
Data for deaths were obtained from the Western Australian Mortality Database. This database is derived from death notifications received by the Western Australian Registrar General’s Office. The computerised information is coded and collated by the Australian Bureau of Statistics, who then forward a copy to the Health Department of Western Australia for integration into the Western Australian Mortality Database.

Only one code is specified on death records as the main cause of death. For most death causes, this code consists of an ICD-9 diagnostic code. For injuries, however, the external cause of death (ie. E code) is specified rather than the diagnosis. For example, if a cyclist dies from intracranial injuries after colliding with a motor vehicle, the recorded cause of death will identify the motor vehicle collision (eg. 813.6) rather than the intracranial injuries. Thus, it is not possible to determine which body part was injured nor the nature of the injury from the cause of death code.

2.2.3 Population Data
In rate calculations, population data (ie. denominators) were obtained from estimates calculated by the Health Department of Western Australia. They are derived from census data collected by the Australian Bureau of Statistics (ABS Catalogue 3203.5). As previously mentioned, the size of the WA cyclist population should ideally have been utilised to examine rate trends. However, as reliable bicycle usage data were not available for the period of interest, overall population figures were used instead. This may be a source of bias if the population of cyclists is changing at a different rate to the State population.

2.3 Analysis

Only the principal reason for hospitalisation and the main cause of death were analysed.

Age-specific rates and age-standardised rates (ASR) were calculated using the Health Statistics Calculator, a software package developed by the Health Department of Western Australia.

All rates were calculated per 100,000 person-years. Where appropriate, the 95% confidence intervals (CI) of age-standardised rates are given in brackets. Where the actual number of deaths or hospitalisations is small, the resultant rate is imprecise and should be quoted with caution.

Due to the low frequency of deaths from bicycle crashes in each period, detailed analysis of mortality data was restricted to the number of deaths by age, sex, place of residence and crash type.

© Health Department of Western Australia. All rights reserved.
Last Updated: Wednesday, 21 October 1998 14:25