Supplementary material from "Suicides on the Austrian railway network: hotspot analysis and effect of proximity to psychiatric institutions"

We model suicide locations on the network using a smoothed inhomogeneous Poisson process and validate it using randomization tests. We find that the density of psychiatric beds is a significant predictor of railway suicide. Further predictors are population density, multitrack structure and—less consistently—spatial socio-economic factors including total suicide rates.

We evaluate the model for the identified hotspots and show that the actual influence of these variables differs across individual hotspots. This analysis provides important information for suicide prevention research and practice. We recommend structural separation of railway tracks from nearby psychiatric institutions to prevent railway suicide.