Background On urban roads, road users interact in a highly complex environment. Few previous multivariable studies investigated the association between the built urban environment (including the road, roadside and human activity) and crashes. Roadside design and amenities and facilities that impact road user movements were rarely considered. This research aimed to identify characteristics of the built urban environment associated with the frequency of pedestrian-vehicle casualty crashes (PVC) on shopping strips in metropolitan Melbourne, Australia.

Methods The study had a cross-sectional design. Strip shopping arterial road segments in Melbourne, Australia, were identified (n = 142). Data were collected for a broad range of potential risk factors relating to the road, roadside and human activity. Analysis using Poisson regression identified factors associated with PVC frequency. Diagnostic tests were conducted to assess model fit.

Results From 2005 to 2009 there were 519 (median = 5.4/km) PVC on the 142 midblock road segments. Aspects of the built environment associated with increased PVC frequency were: road width, unsignalised intersections, medians or traffic islands, midblock pedestrian crossings, roadside development height, off-street parking facilities, establishments with a late night liquor license, railway stations and tertiary education institutions. Lane width and parking clearways were associated with significant reductions in PVC. Speed limits were associated with PVC in a non-linear fashion. There was no association between traffic volume or estimated pedestrian activity and PVC.

Conclusions A range of risk factors were related to PVC on strip shopping road segments including traditionally studied characteristics like road design and less commonly studied aspects such as the roadside and facilities and amenities on the road segment. This study can contribute to the design of PVC countermeasures and better risk assessment tools for urban road segments.