A geographic study into fatal crash rates for local government areas in the state of NSW between 1997 and 2001

When investigating the prevalence of health and disease, epidemiologists routinely conduct geographic studies. Geographic studies alone cannot determine causality, but are often the impetus for further research. Epidemiologists find computer based geographic information systems (GIS) also effective in the monitoring and control of various infectious diseases. Road safety professionals are also using GIS techniques to monitor the performance of the road network, and to assist in the identification of areas that could benefit from engineering interventions. Little work has been published comparing the rates of crashes at a local or micro area level.

The major aim of this study is to compare the geographic variation in fatal crash rates per capita for license holders in NSW. For each Local Government Area (LGA) in NSW, the number of fatal crashes was obtained from the Roads and Traffic Authority’s (RTA’s) Traffic Accident Database. The fatal crash rate per 10,000 license holder years was calculated for each LGA, and mapped using GIS software.

When numbers are small, and rates are seen as imprecise and possibly unreliable measures of risk, the use of a statistical model to generate an alternative indicator of risk seems to be an attractive approach. A Poisson distribution was used to calculate upper and lower confidence limits for the number of fatal crashes in each LGA at a 95% confidence level. Fatal crash rate indicators generated through this process were also mapped.

The three maps were then compared. Geographic clumping of ‘high risk’ areas was evident around major urban population centres when raw fatal crash numbers (by LGA) were mapped. However, when fatal crash rates per license holder were mapped, the pattern of geographic clumping moved from urban to rural and regional areas of the state. A broadly similar pattern was found when mapping the crash rate indicator data.

Number, rate and indicator all provide insights into data interpretation, and these insights are potentially quite different. All of these measures need to be considered carefully by road safety strategists.