|

Traffic Circles’ Associations with motor vehicle, cyclist-MV, and pedestrian-MV crashes

Author(s): Teschke

Slidedeck Presentation:

1B - Teschke

Abstract:

In an earlier study of cycling injuries in Toronto and Vancouver, local street intersections with traffic circles were found to have an 8-fold higher injury risk than those with stop sign control.

An outstanding question is the effect of traffic circles on injuries to people walking and driving. We examine this issue using an Insurance Corporation of BC dataset of 40,626 crashes that occurred at local street intersections in Vancouver from 1996 to 2013. To determine whether traffic circles were associated with the risk of injuries for three types of crashes:

* motor vehicle (MV)
* cyclist-MV
* pedestrian-MV

Relative risks of traffic circles were calculated using the following three comparisons: with vs. without traffic circles; traffic circles vs. matched intersections within one block; and after vs. before traffic circles.

Data on exposure (motor vehicle, cyclist, pedestrian traffic volumes) were sparce for local streets, so could not be used in the analyses. A substitute exposure denominator was used for all analyses: 'intersection-years' (summation, over all intersections with a given comparison status, of the number of years each intersection had that status). Crashes that occurred in the year when a traffic circle was installed or removed were not included in the analyses, so these years were not tallied in intersection-years. For motor vehicle crashes, traffic circles were associated with relative risks of 0.17 to 0.53, resulting in 11 to 14 fewer injuries per year. The risk reduction is consistent with research evidence about the effect of converting stop-controlled intersections to single-lane roundabouts on injuries to motor vehicle occupants, with an average crash modification factor of 0.4 (range 0.12 to 0.65). The relative risks calculated in this analysis are within this range.

For cyclist-MV crashes, traffic circles were associated with relative risks of 2.4 to 4.6, resulting in 10 to 12 more injuries per year. The risk increase is consistent with results of our Vancouver and Toronto study. It found a higher relative risk, 8.0, because it included cyclist-only in addition to cyclist-MV crashes and these comprised half the traffic circle crashes. This suggests that the total increase in traffic-circle-associated cyclist injuries may be 20 to 24 annually.

For pedestrian-MV crashes, traffic circles were associated with relative risks of 0.97 to 1.9, resulting in 0 to 2 more injuries per year. This analysis confirmed an excess risk at traffic circles for people cycling, found a slight increase for people walking and a decrease in risk for motor vehicle occupants similar in magnitude to the increase for those cycling. These results suggest that traffic circles should not be used as a traffic calming measure along local street bike routes. Alternatives such as raised crossings and traffic diversion should be tried to determine whether they are safer for all modes of travel.