The Impact of the ON-S1 Standard on Railway Risk Levels in Australia

The objective of this study was to compare risk levels, based on reported railway occurrences, across Australian states. A secondary aim was to use these numbers to assess the impact of a new reporting system, ON-S1, introduced in 2004, on the calculated risk levels. The Australian nationwide standard, ON-S1, defines how to categorize occurrences and their consequences, but it has not been consistently applied in all states. For example, New South Wales and Victoria use broader definitions of ´serious injury´ than specified in ON-S1. This paper outlines challenges related to ON-SI´s use and the appropriateness of calculating risks based on reported occurrences. Railway occurrence data from five Australian states/territories from 2001 to 2007 were reviewed. Data on fatalities, serious injuries and train kilometres were obtained from the Australian Transport Safety Bureau (ATSB) and The Independent Transport Safety and reliability Regulator´s (ITSRR) safety reports. We used these data to calculate an index of Fatalities and Weighted Injuries (FWI), normalized for train kilometres. The results showed that the average annual risk per million train kilometres for the entire period (2001-2007) was highest in New South Wales (FWI 1.28), followed by Victoria (FWI 0.89), South Australia (FWI 0.77), Queensland (FWI 0.34) and Western Australia (FWI 0.26). Following the introduction of ON-S1, the FWI in New South Wales and Victoria doubled from 2004 to 2006. These trends continued into the first half of 2007. The other states showed a stable or decreasing trend. The highest risk appeared to be in the states with the largest populations containing Australia´s largest cities. The use of train kilometres to normalize occurrence rates may be inappropriate because this measure does not take into account the number of people travelling on each train. The sharp increase in reported occurrences in New South Wales and Victoria may reflect the misuse of ON-S1, because a similar inc- ease was not observed in the other states. The large difference in risk level between states highlights a need for consistent application of the national reporting regime in Australia to enable valid comparisons of occurrence rates between states.