An analysis of 318 fatalities at railway stations

There are approximately 250 fatalities due to suicide each year on the GB network. It is estimated that around 40 per cent of these occur at stations. There are a number of known modes of access to the track (for example... [ view full abstract ]

There are approximately 250 fatalities due to suicide each year on the GB network. It is estimated that around 40 per cent of these occur at stations. There are a number of known modes of access to the track (for example jumping, lying and wandering on the track), requiring different options for prevention.
The RESTRAIL project was an EU collaboration involving 17 partners across twelve different countries in Europe, with the aim of minimising the fatalities on the railway due to suicide and trespass. As part of RESTRAIL, data have been collected in GB on fatalities at 52 stations in the three pilot areas on main lines on three rail routes close to London. This paper presents the results from the analysis of fatalities that have occurred at these stations over a 20 year period.
The collation and analysis of data from the industry Safety Management Information System (SMIS) database are explained. These data have been supplemented where necessary with additional data from project staff in the industry.
The analysis produced descriptive data on the locations of events (e.g. in station, out of station, part of the platform), the mode of access to the track, the times of events, the specific line of incident, summary details of the individuals involved (e.g. age, sex) and impacts of the events for the industry (e.g. delay time and costs). Descriptive data are compared, both within and between rail routes (e.g. frequencies of incidents at stations by year, time of events, locations of events, proportions on fast / slow lines). Text descriptions from the SMIS database have been analysed to identify immediate behaviours, precursor behaviours and contextual factors for each event. Sequences of behaviour and contextual factors have been represented in a series of diagrams, illustrating different phases in the events and links between behaviours and influencing factors. Clustering has been used to determine the relative frequencies of common sequences and anomalous behaviours in the period leading up to an event. These include findings on behaviours that should be of interest to those who are planning preventative strategies (e.g. the numbers of events in which the person crossed one or more lines to meet trains on fast lines). Summary details of the incidents have also been linked to descriptions of specific geographical / design related features that may have influenced behaviour in some locations.
The outputs represent findings for stations on sections of main lines where there are few (if any) crossings and the main point of access to the railway is via the station area. The identification and representation of common sequences of behaviours, precursors and contextual factors associated with the events adds depth to earlier classifications that are available in the literature on the modes of access to the track. The detailed, descriptive findings provide knowledge that is fundamental to the correct specification of recommendations for prevention.