Summary

Findings of Stage 1 of the Austroads-funded development of linkage-based measurement of serious non-fatal road injuries in Australia raised the possibility that the method used in AIHW reports, including those prepared by the National Injury Surveillance Unit (NISU) for the Bureau of Infrastructure, Transport and Regional Economics (BITRE), could be improved by changing the way in which on-road cases are selected from among all records of hospitalised injury cases due to land transport crashes (Harrison et al. 2019).

In the linkage-based project, the cases specified as road injuries (apart from those in which hospital records were linked to crash records) were hospital cases included on the basis of the presence of Place codes that mean the injurious events occurred on a street or highway, rather than on the basis of Traffic codes, which have been the basis in most NISU reports.

The project report showed that the Traffic and Place methods could be expected to produce different results, particularly for cyclist cases, but did not examine the extent of the difference or other characteristics in detail. Pedal cyclists have become increasingly prominent among the cases included in reports of hospitalised road injuries in Australia.

This technical report examines the effects of using the Traffic and Place approaches to specify pedal cyclist road injury cases when using data from the National Hospital Morbidity Database (NHMD), which have been coded according to the Australian clinical modification of the 10th revision of International Classification of Diseases (ICD-10-AM) (ACCD 2017).

Overall, rates of hospitalised injury of cyclists according to the Traffic model were 1.2 to 1.3 times as high as rates for the Place model over the period from 2000–01 to 2017–18.

The Traffic and Place methods gave similar estimates of on-road cyclist cases for older adults; cases with a specified counterpart in collision; and cases that resulted in high threat to life injury, irrespective of age. Estimates were also similar for cases with hospital stays of longer than 2 days.

The Traffic method gave substantially higher estimates of on-road cases than the Place method for cases involving children. For children aged 0–4, rates using the Traffic model were 1.6 to 3.6 times as high as rates using the Place model, while for those aged 65 and rule over rates were similar for both models.

The Traffic method gave substantially higher estimates of on-road cases than the Place method for cases with no specified counterpart in collision.

Characteristics of the data tend to confirm a view that application of an ICD-10 coding rule accounts for much of the difference. The rule requires coders to use a Traffic code both for cases where the record indicates that the case occurred on-road, and for cases where the record is silent about place. The rule does not apply to Place coding. Coders following the rule would apply it to cases in which limited documentation of external cause left the place unknown (for example, ‘Fell from pushbike’), a circumstance that is perhaps particularly likely to occur in cases that did not involve collision with a motor vehicle, did not result in a severe injury, or in which the cyclist was a child.

The Traffic method overestimates on-road cyclist cases. The Place method should not overestimate on-road cases but has the weakness that it does not include the truly on-road cases that can expected to be among the cases for which place is unspecified. Ways to improve measurement in future are proposed.