Analysis of police narratives to identify common themes in bike crash hotspots

Cycling provides a wide range of health benefits and one way to maximize these is to create a safe cycling environment. This is challenging without ongoing surveillance and high quality data. The aim of our study was to use police narrative data to identify factors (not regularly captured in surveillance systems) that may have resulted in similar outcomes within street segments over time. We chose to study clusters of small geographic sections in order to inform targeted changes to the built environment. We mapped police-reported bicycle crashes in the City of Boston and then identified hotspots (defined as areas where crashes occurred more than five times from 2009 to 2012 and at the same location). We then reviewed in detail the narrative reports associated with those hotspots. Finally, in order to describe the bicycle crashes more generally at the city level, we used a combination of hand coding and computer programming to indicate the presence of more commonly collected crash data (e.g., driving under the influence). We found 1806 bike crashes distributed somewhat evenly throughout the the city, yet occurring on 8% of its 18,155 street segments. We reviewed 63 narratives for 7 crash-involved segments (n=7) that had 5 or more crashes. Nearly all hotspot reviews revealed substantive patterns. For instance, on separate occasions 8 cyclists fell within a 500-foot area after getting caught in the train tracks. All were moderately injured. A focused review of police narratives could inform effective changes to the built environment and prevent future injuries.