The 87 cyclists injured in track-involved crashes in this study, recruited at three Toronto emergency departments over an 18-month period, appear to comprise the largest case series involving streetcar and train tracks reported to date. Three other studies identified cases in a single hospital over a similar time period: 41 emergency department cases in Sheffield [3]; ten hospitalized cases in Amsterdam [9]; and five emergency department cases (all e-bike users) in Bern [5]. A Dutch study in 13 hospitals reported on four emergency department cases [14]. As in our study, the dominant scenario for track crashes in most European studies was bike tires being caught in the flangeway [3, 5, 9]. The exception was that all four Dutch cases involved bicycle wheels being deflected by tram rails [14]. Two European studies described the types and severity of injuries (dominantly fractures and about a quarter of cases admitted to hospital) [3, 9]. We did not gather data on the types of injury, but we examined some aspects of injury severity in an earlier analysis [15]. We did not find greater severity (e.g., transport by ambulance, hospital admission) among cyclists injured at sites with streetcar or train tracks.
Personal characteristics
We examined personal, trip and infrastructure characteristics related to track-involved crashes vs. other crash circumstances. Females were over-represented in track crashes (60 % female) compared to other crashes (37 %) and compared to the Toronto cycling population (34 %) [16]. The European studies did not provide comparisons, but reported that the majority of those injured on tracks were male [3, 5, 9]. We found that younger adults (ages 20 to 39, 67 %) were also somewhat over-represented in the track-involved crashes but not significantly so compared to other crashes (62 %). The injured cyclists in our study appear to have been younger than Toronto cyclists in general, based on a report that used different age categories (ages 15 to 34, 37 %), perhaps because our study area included large universities, many of whose students commute by bike [16].
Inexperience and less frequent cycling were associated with track-involved crashes in our study. The Sheffield study began immediately after their first tram line became operational and they found that cycling injuries peaked 3 to 6 months later, then declined by about 50 % [3]. They attributed this to media attention to the issue, supporting the idea that knowledge related to tracks could be helpful. Bike shop personnel contacted in this study all felt that the best protection for cyclists was to know how to behave near tracks, including being alert and crossing the tracks at a perpendicular angle. Similar guidance is provided in an Ontario government cycling skills guide, and adds waiting for breaks in traffic and potentially dismounting to cross tracks [17]. Since left turns can make perpendicular track crossing difficult (especially where there are complicated track patterns, Fig. 2e) and resulted in much higher odds of track-involved crashes, education materials could also encourage two-stage left turns.
Our results provide some support for the idea that increased knowledge or maneuvering skill may help, given that certain demographic groups were over-represented in track-involved crashes (e.g., less frequent cyclists, women), however a number of factors suggest education may not make a great difference. Those crashing on tracks were not especially inexperienced (average cycling frequency of 123 trips per year). Many of the crashes resulted from sudden maneuvers to avoid collisions with motor vehicles, other cyclists and pedestrians, situations that did not allow prior knowledge to be used as planned. Some cyclists (children, people with certain disabilities or who do not speak English) may not be reached by or be able to implement guidance about tracks. Finally, information about how to ride near tracks is long-standing and common in Toronto, yet the injury toll is very high. These caveats underscore the need for other approaches.
Bike tire characteristics
Bike shop personnel reported that some cyclists request tires wide enough not to be caught in track flangeways. Our analyses showed that bike type was associated with whether injury circumstances were track-involved or not. Bike types more frequently in track-involved crashes had either consistently narrow tire widths (racing, single speed) or wide ranges of tire widths (hybrid, city) in the bike shop survey. Over half the tires on commonly sold bicycles were so narrow that they would fit in any of the track flangeways in Toronto. Although bike shop staff thought that only fat bike tires would be guaranteed not to be caught in the flangeways, tires of ~ 50 mm or greater on cruiser, comfort, and bike share bikes may reduce the likelihood of being caught in many, perhaps most, flangeways and are worthy of further study.
Route characteristics
Route type was associated with track-involved crashes. On major streets with no bike infrastructure, it mattered whether there were parked cars or not (Fig. 2). These two route types have similar presence of streetcar or train tracks [7], but those without parked cars had less than half the odds of track-involved crashes. The absence of car parking provides cyclists with more room to maneuver and avoid track crashes when something unexpected takes place in front of them (Table 1). Removing car parking on streets with streetcar lines would improve conditions for cycling, especially if the space freed up were used for cycle tracks (as discussed below). Painted bike lanes, residential streets, and sidewalks or multiuse paths all had considerably lower odds of track-involved crashes than major streets with no bike infrastructure. This was almost certainly because these route types were much less likely to have streetcar or train tracks [7].
In the Netherlands, cyclists on major streets are typically provided cycle tracks (also called physically protected, segregated or separated bike lanes) and on-street tram lines typically have their own rights of way [14, 18]. This may account for the low numbers of tram-related crashes observed in the Dutch study [14]. Our injury study showed that cycle tracks greatly reduced injury risk to bicyclists [4, 7], but at the time of the study all examples of this infrastructure were in Vancouver, not Toronto. One Toronto streetcar line had its own right of way during the study period (Fig. 2) and all train lines did. We did a post-hoc check of whether any of the track-involved crashes were along these lines. None of track crashes between intersections were along them, but some were at their intersections (where there is no separation). Even if cycle tracks or designated rail rights of way would prevent only crashes that are not at intersections, track-involved crashes would be substantially reduced, since most (68 %) were not at intersections.
Left turns at intersections were highly overrepresented in track-involved crashes. This problem could also be addressed with Dutch-style infrastructure, often called “protected intersections”. Such intersections commonly feature corner islands that direct cyclists coming from cycle tracks to make two-stage left turns, as pedestrians do [19]. This would make it much easier to cross tracks at right angles, but could add long delays at intersections unless signal timing is optimized for cyclists and pedestrians, as in the Netherlands.
Protected intersections, cycle tracks and designated rail rights of way all follow the Swedish “Vision Zero” transport safety principle: acknowledging the inevitability of human error and providing route designs that minimize its consequences [20]. This vision aims to eliminate deaths and serious injuries related to transportation and is beginning to be adopted by other jurisdictions in Europe and North America.
Strengths and limitations
This study benefitted from a large case series of track-involved crashes, a comparison group with other crash circumstances, and systematic data on crash circumstances, personal and trip characteristics, and route infrastructure at the crash sites. The mixed methods approach also collected data about advice provided to cyclists by bike shop personnel, widths of commonly sold bike tires, and engineering specifications of system rails and flangeways to provide a broader understanding of the problem and potential solutions.
Additional data would be helpful in future studies. We did not request data from the injured cyclists about their tire widths. This would be worthwhile to collect, so widths of tires involved in crashes can be compared to flangeway widths and risk related to tire width can be directly determined. Other characteristics (tire pressure, presence of tire knobs, weight of the cyclist) may alter the effective tire width and should be measured to see if they change the tire size needed to avoid being caught in flangeways. Direct measurements of flangeway widths throughout the rail system would be useful, though taking measurements in situ would be a dangerous endeavor. Similarly, field tests of different tire widths with bicycling track interaction maneuvers would be informative but risky to participants. In cities with streetcar or tram systems, it would be interesting to survey cyclists to see if they know how to reduce their individual risk of track crashes, and to survey planners and engineers to see whether they are familiar with design measures to reduce population risk of track crashes.
This study was conducted in one city in North America. Research conducted in other areas of the world with different cycling infrastructure, streetcar or tram infrastructure, and bicycle types would help determine whether these influence risk. Comparisons between cities and countries would be a great way to discover best practices. Unfortunately, such comparisons are difficult because the most common coding system for traffic injuries, the World Health Organization’s International Classification of Diseases [21], provides coding for collisions with a streetcar or train, but codes collisions involving tracks in a broad category of unspecified “stationary objects”. Crash and injury reporting systems that provide sufficient specificity to identify track-related crashes would allow administrative data to be used to tally these events, a crucial first step in understanding their impact [14].