Abstract

US municipalities are introducing bicycle lanes into their roadway infrastructure to promote bicycle use, increase roadway safety and improve public health. The aim of this study was to identify specific locations where bicycle lanes, if created, could most effectively reduce crash rates. Previous research has found that bike lanes reduce crash incidence, but a lack of comprehensive (e.g., city-wide) bicycle traffic flow data has limited researchers’ ability to assess relationships at high spatial resolution and for all locations within a study region. Our unit of analysis was the street segment in Philadelphia, PA (n=37,673). We used Bayesian conditional autoregressive logit models to relate the odds that a reportable bicycle injury crash occurred on a street segment between 2011 and 2014 to characteristics of the street (street class, vehicle and bicycle traffic volume, one way, presence of streetcar tracks and bike lanes) and adjacent intersections (number of exits, stop type). Model 1 identified that crashes are 30% more likely to occur on segments with bicycle lanes (OR=1.30; 95 CI: 1.18, 1.51), perhaps due to unmeasured bicycle traffic through these locations. Model 2 included interaction terms between bicycle lanes and the other street and intersection characteristics, thereby addressing the problem of unknown bicycle traffic flows. In this model, bicycle lanes were associated with 45% reduced crash odds in streets segments adjacent to intersections with 4 exits (OR=0.55; 95% CI: 0.38; 0.80) and with 37% reduced crash odds in streets adjacent to intersections with one- or two-way stop signs (OR=0.63; 95% CI: 0.46; 0.88). The effectiveness of bicycle lanes appears to depend most substantially on the configuration of the adjacent intersections. Results also suggest locations at which the greatest absolute reduction in bicycle crash odds could occur by installing bicycle lanes.