Toronto bicycle commuter safety rates

https://doi.org/10.1016/S0001-4575(99)00028-7Get rights and content

Abstract

This analysis uses data from a survey of Toronto commuter cyclists that collected information regarding accident history as well as regular commute route to work or school. By relating the route information of the 1196 respondents to facility attributes in a Geographic Information System (GIS), defensible estimates of travel exposure on roads, off-road paths and sidewalks were developed. The rate of collision on off-road paths and sidewalks was lower than for roads. The relative rates for falls and injuries suggest these events are least common on-road followed by off-road paths, and finally most common on sidewalks. The rate of major injuries, an injury that required medical attention, was greatest on sidewalks and the difference between paths and sidewalks was negligible. These rates suggest a need for detailed analysis of sidewalk and off-road path bicycle safety. The absolute event rates per bicycle kilometer were found to be between 26 and 68 times higher than similar rates for automobile travel, re-confirming the urgent bicycle safety crisis. Examination of rates for sub-groups of cyclists suggest that experience is an important factor in bicycle safety. The same survey conducted in Ottawa, Canada found event rates much lower than Toronto. This result may confirm urban form, traffic levels and attitude do affect bicycle safety. The analysis also demonstrates a successful method to quantify bicycle travel exposure information and should be considered for further use as complement to other existing techniques.

Introduction

In order to properly plan for the bicycle as an urban mode of transportation, a measure of the relative safety of on- and off-road bicycle travel is needed. The debate over the relative merits of provision of special bicycle infrastructure, such as paths, has raged for decades. There are contradictory results among the few existing studies, some finding special bicycle infrastructure safer than roads while others finding the opposite. Until recently, only one comprehensive North American study with a very large number of respondents over more than just a single travel corridor produced accident rates based on estimates of travel exposure for different types of infrastructure. These results suggested that it was less safe to travel on bike paths than on roads (Kaplan, 1976). Recent research (Moritz, 1997) using an approach similar to Kaplan’s nation-wide survey of bicycle club members found accident rates highest on sidewalks, followed by paths and then various categories of roads. The previous application of the analysis method used in this paper (Aultman-Hall and Hall, 1998a) to a sample of Ottawa, Canada commuter cyclists found the same event rate pattern: highest on sidewalks followed by paths1 and then roads. All three sets of results are in direct contradiction with the subjective perception of many bicyclists and non-bicyclists who feel cycling away from traffic is safer (Goldsmith, 1992, Badgett et al., 1993). Furthermore, the safety concerns are disturbing in that both European and North American experience indicates that off-road facilities increase cycling levels which is good from a traffic congestion, health and economic point of view. Given these contradictions there is a need for further comprehensive quantitative analysis to measure actual bicycle event rates. This paper presents results of a bicycle safety survey of bicycle commuters in downtown Toronto, Canada. The objective is to present the Toronto on and off-road event rates as well and to compare them with earlier results from Ottawa (Aultman-Hall and Hall, 1998a).

Before presenting the details of the analysis and findings, the next section of this paper reviews the survey methodology and the characteristics of the study area. A more comprehensive review of background literature was presented in (Aultman-Hall and Hall, 1998a). The subsequent three sections of the paper each have two subsections, one dealing with methodology and procedures, and the second reporting the results. These subsections describe the estimation of travel exposure; the relative event rates for travel on roads, off-road paths and sidewalks; and finally the relative event rates for different subgroups of cyclists. The final section of the paper presents possible interpretations for the results and points to the direction of further work.

Section snippets

The survey data and characteristics of the study area

There are two key limitations that hinder bicycle safety analysis: lack of complete incident databases; and lack of travel exposure information. For this study, self-reported falls and collisions are used from a survey that asked cyclists to indicate their accident history. This is intended to compliment the more typical use of police and emergency room accident databases. While self-reported databases by definition do not contain fatalities or some serious injuries, they do contain information

Procedure

The network of roads, off-road paths, and intersections for the Toronto study area was represented as a system of links and nodes in an Arc/Info GIS coverage. The routes traced on the maps by the cyclists were entered into this network coverage. The links in the GIS coverage have an associated attribute table indicating the type of link as a path or road. Each individual’s route was related to the coverage’s attribute tables to get the exact length of travel on each type of infrastructure in

Analysis procedure

The event rate analysis procedure consisted of two parts: calculation of exposure and event totals adjusted for confounding personal variables; and statistical tests for relative event rates.

There are four types of events of interest in this study: collisions, falls, injuries and major injuries (this final category is a subset of the previous one). Due to the rare nature of all events, this initial analysis considers aggregate event rates. This has implications not just for how event rates are

Analysis procedure

The degree to which a sample can be subdivided is determined by the size of the sample and the number of events reported. Given the overall event sample size (at best only 300), only one-way analysis is considered in this section. The Hauer hypothesis test described earlier is used to compare the event totals given different travel exposure levels for different subgroups of cyclists, in order to determine if event rates per travel distance differ significantly.

Two slightly different procedures

Concluding discussion

This study has found statistically significant differences between the collision, fall and injury rates for bicycle commuting on-road, off-road and on sidewalks in the Toronto study area. In general, these relative rates suggest it is safest to travel on-road followed by off-road paths, and finally, least safe on sidewalks. While the same analysis undertaken in Ottawa resulted in the same pattern of relative rates, the magnitudes were different. The rates per distance are all higher in Toronto

Acknowledgements

The authors wish to thank the Ontario Ministry of Transportation and the Natural Sciences and Engineering Research Council of Canada for financial support. The provision of data from the Regional Municipality of Metropolitan Toronto and the City of Toronto was essential to the undertaking of this research. Fred L. Hall of McMaster University made significant contributions to the development of the survey and analysis methods used in this paper.

References (10)

  • L. Aultman-Hall et al.

    Ottawa-Carleton on- and off-road cyclist accident rates

    Accid. Anal. Prev.

    (1998)
  • Aultman-Hall, L., Hall, F.L., 1998b. Research design insights from a survey of urban bicycle commuters. Transportation...
  • Aultman-Hall, L., Adams, M.F., 1998. Sidewalk bicycling safety issues. Transportation Research Record...
  • Badgett, S.I., Niemeier, D.A., Rutherford, G.S., 1993. Bicycle commuting deterrents and incentives: a survey of...
  • J. Forester

    Bicycle Transportation—A Handbook for Cycling Transportation Engineers

    (1994)
There are more references available in the full text version of this article.

Cited by (0)

View full text