Background
Previous research has shown an inconsistent relationship between the spatial distribution of hospital treated self-harm and area-level factors such as deprivation and social fragmentation. However, many of these studies have been confined to urban centres, with few focusing on rural settings and even fewer studies carried out at a national level. Furthermore, no previous research has investigated if travel time to hospital services can explain the area-level variation in the incidence of hospital treated self-harm.

Methods
From 2009 to 2011, the Irish National Registry of Deliberate Self Harm collected data on self-harm presentations to all hospital emergency departments in the country. The Registry uses standard methods of case ascertainment and also geocodes patient addresses to small area geographical level. Negative binomial regression was used to explore the ecological relationship between area-level self-harm rates and various area-level factors.

Results
Deprivation, social fragmentation and population density had a positive linear association with self-harm, with deprivation having the strongest independent effect. Furthermore, self-harm incidence was found to be elevated in areas that had shorter journey times to hospital. However, while this association became attenuated after controlling for other area-level factors it still remained statistically significant. A subgroup analysis examining the effect of travel time on specific methods of self-harm, found that this effect was most marked for self-harm acts involving minor self-cutting.

Conclusions
Self-harm incidence was influenced by proximity to hospital services, population density and social fragmentation; however, the strongest area-level predictor of self-harm was deprivation.