UBC Faculty Research and Publications

The impact of transportation infrastructure on bicycling injuries and crashes: a review of the literature Reynolds, Conor C; Anne Harris, M; Teschke, Kay; Cripton, Peter A; Winters, Meghan Oct 21, 2009

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12940_2009_Article_266.pdf [ 643.12kB ]
JSON: 52383-1.0074677.json
JSON-LD: 52383-1.0074677-ld.json
RDF/XML (Pretty): 52383-1.0074677-rdf.xml
RDF/JSON: 52383-1.0074677-rdf.json
Turtle: 52383-1.0074677-turtle.txt
N-Triples: 52383-1.0074677-rdf-ntriples.txt
Original Record: 52383-1.0074677-source.json
Full Text

Full Text

ralssBioMed CentEnvironmental HealthOpen AcceReviewThe impact of transportation infrastructure on bicycling injuries and crashes: a review of the literatureConor CO Reynolds*1, M Anne Harris2, Kay Teschke2,3, Peter A Cripton4 and Meghan Winters2Address: 1Institute for Resources, Environment and Sustainability, University of British Columbia (UBC), 2202 Main Mall, Vancouver, BC, V6T 1Z4, Canada, 2School of Population and Public Health, UBC, James Mather Building, 5804 Fairview Avenue, Vancouver, BC V6T, 1Z3, Canada, 3School of Environmental Health, 2206 East Mall, UBC, Vancouver, BC, V6T 1Z3, Canada and 4Department of Mechanical Engineering, UBC, 2054-6250 Applied Science Lane, Vancouver, BC, V6T 1Z4, CanadaEmail: Conor CO Reynolds* - c.reynolds@ires.ubc.ca; M Anne Harris - aharris7@interchange.ubc.ca; Kay Teschke - kay.teschke@ubc.ca; Peter A Cripton - cripton@mech.ubc.ca; Meghan Winters - mwinters@interchange.ubc.ca* Corresponding author    AbstractBackground: Bicycling has the potential to improve fitness, diminish obesity, and reduce noise, airpollution, and greenhouse gases associated with travel. However, bicyclists incur a higher risk of injuriesrequiring hospitalization than motor vehicle occupants. Therefore, understanding ways of making bicyclingsafer and increasing rates of bicycling are important to improving population health. There is a growingbody of research examining transportation infrastructure and the risk of injury to bicyclists.Methods: We reviewed studies of the impact of transportation infrastructure on bicyclist safety. Theresults were tabulated within two categories of infrastructure, namely that at intersections (e.g.roundabouts, traffic lights) or between intersections on "straightaways" (e.g. bike lanes or paths). To assesssafety, studies examining the following outcomes were included: injuries; injury severity; and crashes(collisions and/or falls).Results: The literature to date on transportation infrastructure and cyclist safety is limited by theincomplete range of facilities studied and difficulties in controlling for exposure to risk. However, evidencefrom the 23 papers reviewed (eight that examined intersections and 15 that examined straightaways)suggests that infrastructure influences injury and crash risk. Intersection studies focused mainly onroundabouts. They found that multi-lane roundabouts can significantly increase risk to bicyclists unless aseparated cycle track is included in the design. Studies of straightaways grouped facilities into fewcategories, such that facilities with potentially different risks may have been classified within a singlecategory. Results to date suggest that sidewalks and multi-use trails pose the highest risk, major roads aremore hazardous than minor roads, and the presence of bicycle facilities (e.g. on-road bike routes, on-roadmarked bike lanes, and off-road bike paths) was associated with the lowest risk.Conclusion: Evidence is beginning to accumulate that purpose-built bicycle-specific facilities reducecrashes and injuries among cyclists, providing the basis for initial transportation engineering guidelines forcyclist safety. Street lighting, paved surfaces, and low-angled grades are additional factors that appear toimprove cyclist safety. Future research examining a greater variety of infrastructure would allowPublished: 21 October 2009Environmental Health 2009, 8:47 doi:10.1186/1476-069X-8-47Received: 10 July 2009Accepted: 21 October 2009This article is available from: http://www.ehjournal.net/content/8/1/47© 2009 Reynolds et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 19(page number not for citation purposes)development of more detailed guidelines.Environmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47BackgroundBicycling is an active mode of transportation that inte-grates physical activity into daily life. The bicycle is anattractive alternative to the automobile as an urban modeof transport. Cycling is associated with a range of individ-ual and public health benefits such as improved physicaland mental health, decreased obesity, and reduced risk ofcardiovascular and other diseases [1-6], as well as ancil-lary benefits such as reduced emissions of noise, air pol-lutants and greenhouse gases [7,8]. There are significanteconomic costs of physical inactivity [9], and benefit-costanalyses suggest that the benefits of increased cycling areworth approximately four to five times the costs of invest-ing in new cycling infrastructure [10,11]. These potentialbenefits suggest that it is important to increase the use ofthe bicycle as a mode of active transportation.It is clear that the health benefits of cycling are significant,and at this point there is no reason to assume that healthrisks outweigh those benefits. However, a full publichealth understanding requires that attention be paid notonly to long-term population health and environmentalbenefits of bicycling, but also to the factors that influencerisk of injury and fatality. Bicyclists are vulnerable becausethey must frequently share the same infrastructure withmotorized vehicles, and yet bicycles offer their users nophysical protection in the event of a crash. In addition, themass of a typical automobile is at least an order of magni-tude greater than a bicycle plus its rider, and motorizedvehicles have top speeds that are considerably faster thanbicycles. As a result, bicycle riders who are involved in acrash are exposed to a much higher risk of injury com-pared to motor vehicle users (with the exception ofmotorcycle riders).To date, most studies of cycling safety - especially in NorthAmerica - have emphasized helmet design, regulation,and implementation to mitigate the severity of cyclinginjuries when a crash occurs [12,13]. This is particularlytrue for children [14,15]. In many North American juris-dictions children who cycle (and sometimes also adultcyclists) are required by law to use helmets, although thisis not the case in most European countries. While helmetsare effective in reducing the severity of head injuries, theydo not address impacts to other parts of the body [16,17].More importantly, they do not prevent incidents fromoccurring in the first place [18], and legislating their usemay even discourage cycling [19].The built environment has been implicated as an impor-tant determinant of bicycling rate [20-23], but these rela-tionships are complex and a positive correlation has notalways been found [24]. It is equally important to under-because there may be an opportunity to prevent injuriesby modifying transportation infrastructure. Infrastructureimprovement meets several important conditions for suc-cessful injury prevention measures: (a) it is populationbased, rather than requiring initiative on the part of theindividual; (b) it is passive, rather than requiring activeparticipation; and (c) it is accomplished with a singleaction, rather than requiring repeated reinforcement [18].In this paper we review the evidence on how differenttypes of transportation infrastructure affect bicyclists'safety. This paper is organized as follows: first we providean overview of bicycling safety and ridership. Next weoffer definitions of, and alternative terminology for, thetransportation infrastructure used by cyclists that might beexpected to influence their safety (Table 1). We describeour literature search methodology and the inclusion andexclusion criteria, and present the results of the search intwo detailed tables. Table 2 describes studies that assessthe safety of intersections for cyclists, and Table 3describes studies related to straightaways (i.e. roads, lanes,paths). We conclude by discussing the findings of thisreview, critiquing the methodological approaches used,and offering recommendations for future research.Ridership and SafetyNorth Americans remain less likely than Europeans tochoose bicycle transport for either short or long trips. Inpart, this may be due to differences in urban formbetween North American and European cities, particularlydensity and interconnectedness [25], but perceived andactual injury risks are also important.Data on the share of trips made by bicycle are not oftendirectly comparable between jurisdictions owing to differ-ences in the survey methods employed (e.g. samplingscheme, definition of a trip, etc.), but comparisons aretypically justified by the inability of these methodologicaldisparities to explain the substantial difference observed(e.g. about 1% of trips are made by cycling in North Amer-ica vs. an estimated 10% of trips in Switzerland, Germany,Austria, Sweden, Finland, and Belgium (Flanders), andmore than 20% of trips in Denmark and the Netherlands)[26]. Along with these lower cycling rates, there is also ahigher risk of injury associated with cycling in NorthAmerica: an analysis of traffic injuries indicated a two tothree fold higher risk of death and an eight to 30 foldhigher risk of injury while cycling in the United States vs.Holland and Germany, using either of the traditionaltransportation denominators: per trip or per kilometertraveled [27]. While these comparisons underscorecycling injury risks, they also provide reason for opti-mism. If cycling is safer in European cities, it can be madePage 2 of 19(page number not for citation purposes)stand how the built environment affects bicycling safety safer in North America.Environmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47Page 3 of 19(page number not for citation purposes)Table 1: Key terminology for describing transportation infrastructure used by cyclistsTerm DescriptionSTRAIGHTAWAYSOn-road cycling/vehicular cycling When bicyclists ride on a roadway designed primarily for motor vehicles.Wide curb lane The outer (curbside) lane of a paved multi-lane road is wider than the standard width and can accommodate cyclists, although there may not be signs indicating this.Sharrows *(suggested cycle lane†)Symbols painted on the paved roadway indicating that bicycles can share the lane with motor vehicles. They are sometimes used on roads with high cyclist traffic that don't have enough width to accommodate a bike lane.Bike route A paved residential or local road that is signed as being a "bike route", and may have cyclist-friendly crossings at major roads, such as traffic signals with push-buttons that are easily operated by cyclists.Bike lane(marked cycle lane†)Part of the paved roadway marked with painted lines or a colored surface, to designate that it is reserved exclusively for cyclists. Bike lanes may terminate before an intersection, or continue through it.Cycle track(separated cycle lane†)Paved lane, exclusively for bicycle use, next to a major city street or roundabout, but separated by a curb or other physical barrier.Bike path Off-road paved or unpaved path or trail, for bicycles only.Multi-use path Off-road paved or unpaved path or trail, shared with other non-motorized users (e.g. pedestrians, runners, or in-line skaters).Sidewalk Off-road paved walkway for pedestrian use, located by the side of road; known as "pavement" in some parts of the world (e.g. UK and Ireland).Speed bumps/humps * Raised ridge across the road designed to slow motor vehicle traffic ("traffic calming"), particularly in residential areas. Speed humps are easier than speed bumps for cyclists to ride over because they are less steep-sided and more broad.INTERSECTIONSIntersections Where two or more roads either meet or cross at the same level.Junctions * May be road intersections, but the term is usually used to refer to the point where a laneway, path, or driveway meets a road.Roundabout Intersection of arterial streets with a central circle of sufficient diameter that the road curvature accommodates all road vehicles, including trucks and buses. Roundabouts usually have splitter islands on the approaches, sidewalks around the edges, and crosswalks across the approaches set back from the intersection. Daniels et al. provide diagrams of different types of cycle facilities on roundabouts in the Netherlands [57]. Generally, entering traffic yields to traffic already in the intersection.Traffic circle/rotary traffic island * Raised concrete circles placed in the centre of minor street intersections; there are no splitter islands and the design vehicle is a passenger car.Bicycle crossing Distinct road crossings for cyclists that are sometimes raised or colored, and may have cyclist-operated traffic signals.Bicycle box/advanced cycle stop line * A right-angle extension to a bike lane at the head of an intersection, which allows cyclists to wait at the head of the traffic queue on a red traffic signal and then proceed through the intersection ahead of motor vehicle traffic on green.Traffic diverter * Bike-permeable barriers that require motor vehicle traffic to turn instead of traveling straight ahead through an intersection, or that prevent motor vehicles from entering a street.* These types of infrastructure were not investigated in any of the studies identified for this review.† Terminology used in the "European Cycling Lexicon" (published by the European Economic and Social Committee at the Vélocity 2009 conference in Brussels). It gives a list of key cycling terms with corresponding photographs for cyclists and policy makers, in all 23 official European languages. It is freely available to download at: http://www.eesc.europa.eu/sections/ten/european-cycling-lexiconEnvironmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47There are clearly bicycling safety and popularity "gaps"between (and within) Europe and North America [28]. Inaddition, there is an important safety gap between cyclistsand other transport modes: estimates from both conti-nents suggest that cyclists are seven to 70 times morelikely to be injured, per trip or per kilometer traveled, thancar occupants [27,29]. It is likely that public perception ofa lack of safety acts as a deterrent to cyclists in NorthAmerica: in surveys asking about factors that affect thechoice of cycling as a mode of transportation, concernabout safety is one of the most frequently cited deterrents[[30-34], and Winters M, Davidson G, Kao D, Teschke K:Motivators and deterrents of bicycling: factors influencingdecisions to ride, submitted]. For example, in a survey ofadults in the Vancouver metropolitan area, the followingwere among the top deterrents: the risk of injury from car-bike collisions; the risk from motorists who don't knowhow to drive safely near bicycles; motorized vehicles driv-ing faster than 50 km/hr; and streets with a lot of car, bus,and truck traffic [33]. The good news is that there is evi-dence that perceived safety improvements in bicycle trans-portation have an aggregate elasticity value greater thanone (i.e. a 10% increase in perceived safety results ingreater than 10% increase in the share of people commut-ing by bicycle) [32].Increased ridership rates may result in improved safety forcyclists: injury rates have been shown to decrease withincreased cycling rates. This principle of "safety in num-bers" is supported by studies of injury and ridership pat-terns in California, Australia, and Europe, as well asbetween cities and within cities over time [35-38]. Thereare a number of potential explanations. Motor vehicledrivers may not expect cyclists when there are few of themon the roads, and thus make so-called "looked-but-failed-to-see" errors that can result in collisions [39]. Whenmotorists and cyclists are unaccustomed to sharing theroad, both parties may hold incorrect assumptions aboutwhat the other party will do [40]. Increased cycling ratesmay mean that more motorists also use bicycles as a modeof transport, making motorists more attuned to cyclistsand their movements, and encouraging them to drive in away that accounts for potential interactions [36]. Finally,a larger cycling population means stronger lobbyingpower for cycling resources.Finally, it is worth considering long-term temporal trendsin motor vehicle injuries. The injury rate from motor-vehi-cle crashes has steadily declined since the 1920s in manyparts of the world, in part attributable to improvements inroad-related infrastructure [41]. This provides reason foroptimism: the risk of injury or death from traffic crashes ismodifiable, and this is likely to extend to the infrastruc-Safety and Infrastructure TerminologySafety terminologyyBicycling safety is usually quantified by measuring one ormore of the following metrics: injuries; crashes; and con-flicts. Injuries may include fatalities and can be classifiedaccording to their type and severity using standardizedmethods such as the World Health Organization's Inter-national Classification of Diseases (ICD) [42] and theAssociation for the Advancement of Automotive Medi-cine's Abbreviated Injury Scale (AIS). Crashes can be clas-sified as either a collision or a fall, where a collision isdefined as an event in which the bicycle hits or is hit byany other object, regardless of fault, and a fall is an event(not caused by a collision) where the bicycle and/or bicy-clist lands on the ground.A conflict is normally defined as an interaction between abicyclist and another road user such that at least one of theparties has to change speed or direction to avoid a colli-sion. Types of conflict examined in bicycling safety studiesinclude avoidance maneuvers at intersections [43-45],bicycle-motor vehicle interactions during passing eventson roads, lanes, or paths [46-49], and "wrong side passingevents" on multi-use paths [50]. Conflict studies mayoffer valuable insights into how cyclists and other roadusers behave during their interactions on various types oftransportation infrastructure. However, it is not possibleto determine whether the safety of the cyclists was com-promised during the conflict events. In addition, the con-flict studies we identified were generally based on a smallnumber of observed events, which were made over a lim-ited time period (usually several hours), and often in asingle geographical location. Therefore, papers that usedconflict as their sole outcome measure have not beenincluded in this review.In the literature that examines traffic-related injuries andcrashes (including many of the papers reviewed here) theword "accident" is frequently used, for example in thephrase "motor vehicle accident". However it has beenargued that the term "accident" implies that the event inquestion has happened entirely by chance, and is there-fore unpredictable and unpreventable [51] as opposed tobeing a result of modifiable risk factors. The editors ofBMJ have even gone as far as to ban the use of the term[52]. We have refrained from using the word "accident" inthis review, instead using the more specific terms "inci-dent", "injury", "crash", "collision" and "fall" as appropri-ate. However, we do indicate if the original study authorsused the word accident to describe the outcome measure.Infrastructure terminologyKey terms that describe transportation infrastructure usedPage 4 of 19(page number not for citation purposes)tural determinants of cycling injuries. by cyclists are defined in Table 1. We have indicated if a/1/47Page 5 of 19(page number not for citation purposes)Table 2: Studies that investigated relationships between bicyclist safety and intersection-related transportation infrastructureReference Location; Infrastructure Study Outcome Analysis Control method Effects observedmporal njury rates ns in the l data that ecrease d. A seven-period" t t included in .8% reduction in bicyclists' crash rate and 30% reduction in injury rate were observed following installation of new roundabouts. Among the 3 styles of roundabouts, those with cycle tracks had the greatest reductions in injuries to cyclists and moped users (90%), compared to those with no bicycle infrastructure (41% reduction) and those with a cycle lane (25% reduction). crashes and ed r ctions based d bicycle At two-lane roundabouts, the observed crashes and injuries were more than twice those expected, whereas at single lane roundabouts there was no difference between expected and observed. Two other factors were associated with lower than expected crashes: single lane roundabouts with a central island radius > 10 m, and bicycle travel on bikeways rather than the roadway of the roundabout intersection.al changes In multiple regression, higher vehicle and cyclist traffic volumes and "drive curve" (a proxy for vehicle speed) were associated with higher numbers of cyclist crashes/year.Environmental Health 2009, 8:47http://www.ehjournal.net/content/8 Design types examined population measures methodROUNDABOUTSSchoon and Van Minnen(1994)The safety of roundabouts in the Netherlands[53]The Netherlands;Observational, before-after interventionRoundabouts vs. other intersection types; and roundabout design features181 intersections before and after implementation of roundaboutsNational database of bicycle and moped injuries and crashes* (529 before, 111 after)Change in crash and injury rates after intervention.Corrected for the tetrends in crash and iacross all intersectioNetherlands: nationashowed a 2 to 13% dover the study periomonth "transitional following roundabouconstruction was nobefore-after analysisBrüde and Larsson (2000)What roundabout design provides the highest possible safety?[54]Sweden;Observational, non-interventionRoundabouts vs. other intersection types72 roundabouts with ≥ 100 cyclists/dayPolice reports of 67 crashes*, 58 of which resulted in injuriesComparison of observed and expected crash counts. Regression analyses to examine factors affecting crash counts and rates.Calculated expectedinjuries using publishprediction models foconventional interseon motor vehicle antraffic volumes.Hels and Orozova-Bekkevold(2007)The effect of roundabout design features on cyclist crash rate[55]Denmark - Odense; Observational, non-interventionRoundabout design features88 roundabouts Police reports and Emergency Department records of 152 injuries*Poisson regression and logistic regression analyses between cyclist injuries (3/year and probability, respectively) and roundabout characteristics: geometry, age, traffic volume (vehicles and cyclists), and location (urban/rural).Adjusted for temporin traffic volume.Environmental Health 2009, 8:47http://www.ehjournal.net/content/8/1/47Page 6 of 19(page number not for citation purposes)Da(20Throutrabicobstu[56omparison group: unchanged onventional intersections near tervention sites to account for emporal trends in safety and egression-to-the-mean (e.g. tersections may have been elected for roundabout onstruction because of higher umbers of crashes).Roundabouts have the effect of increasing risk of crashes resulting in injury at or near the intersection (odds ratio = 1.27). The effect is stronger for intersections inside built-up areas (odds ratio = 1.48).Da(20Injwitrouinflocchathefac[57ame data as Daniels 2008, bove.Roundabouts with cycle lanes had significantly higher risk (odds ratio = 1.93), whereas no increased risks were observed for roundabouts with mixed traffic, separate cycle tracks, or grade-separated paths. Roundabouts with 2 lanes and those replacing signalized intersections also had elevated risks.BICGå(19Mesafraicroneme[58djusted for traffic volume data ollected on 2 intervention treets and 2 unchanged streets.There was an 8% increase in crash frequency in the study area, but bicycle volume on these intervention sections grew by 50% more than unchanged streets - authors conclude that the intervention may have resulted in a safety improvement.Tab astructure (Continued)niels et al. 08)e effects of ndabouts on ffic safety for yclists: An servational dy]Belgium - Flanders;Observational, before-after interventionRoundabouts vs. other intersection types91 intersections before and after implementation of roundabouts (40 inside built-up areas with speed limit of 50 km/h, and 51 in areas with speed limits of 70 or 90 km/h)Police reports of 1060 injuries (411 at roundabouts, 649 at comparison intersections)Effectiveness index = odds ratio for the before-after change in injury rates of the roundabout intersections as compared to the change in injury rates at conventional comparison intersections.Ccintrinscnniels et al. 09)ury crashes h bicyclists at ndabouts: luence of some ation racteristics and  design of cycle ilities]Belgium - Flanders;Observational, before-after interventionRoundabouts vs. other intersection typesSame data as Daniels 2008, above, except only 50 intersections in areas with speed limits of 70 or 90 km/h)Same data as Daniels 2008, above.Effectiveness index as Daniels 2008, above. Regression models to evaluate the roundabout design determinants of the effectiveness index.SaYCLE CROSSINGSrder et al. 98)asuring the ety effect of sed bicycle ssings using a w research thodology]Sweden - Gothenburg;Observational, before-after interventionBicycle crossings (raised above road level by 4-12 cm) vs. other intersection types44 intersections (and 18.7 km of adjacent road sections) before and after implementation of raised bicycle crossingsPolice or hospital reports of 287 crashes* (160 before, 127 after)Calculated unadjusted number of crashes per month after intervention compared to before intervention.Acsle 2: Studies that investigated relationships between bicyclist safety and intersection-related transportation infrEnvironmental Health 2009, 8:47http://www.ehjournal.net/content/8/1/47Page 7 of 19(page number not for citation purposes)Jensen (2008) Safety effects ofblue cycle crossings: a before-after stu[59]son of  injuries es with  (using  random odels).Adjusted for temporal trends in traffic volumes and crashes, based on data from changed and unchanged intersections. Considered regression-to-the-mean, but no adjustment was deemed necessary.Risk of crash/injury depends on number of colored crossings: 1 crossing = 10% reduction for injuries/19% for crashes; 2 crossings = 23%/48% increase; 4 crossings = 60%/139% increase. Authors hypothesize that non-intuitive findings may result from motorist confusion at sites with many crossings.INTERSECTIONWang and Niha(2004)Estimating the riof collisions between bicycleand motor vehicles at signalized intersections[60]isson f crash k: for " motor avel; left-el; and  travel.Adjusted for average bicycle and motor vehicle volume, intersection location, speed limit, visual noise.A higher number of turning lanes and presence of a wide median significantly increased risk of crash during motor vehicle turning maneuvers. Narrower entering approaches and wider medians increased crash risk in certain turning collisions. Increased cycle volumes were associated with lower collision risk with turning vehicles.*These studies u e resulted in injury. We have substituted the words "crash", "collision" and/or "fall" based on o† In Japan, traffic ith pedestrians, not on the road.Table 2: Studie ortation infrastructure (Continued) dyDenmark - Copenhagen;Observational, before-after interventionBicycle crossings (colored blue) vs. other intersection types65 intersections before and after implementation of blue bicycle crossingsPolice reports of 567 injuries (319 before, 248 after); 1,595 collisions (778 before, 817 after)Compariobservedand crashexpectedfixed andeffects m DESIGNn sk s Japan - Tokyo†;Observational, non-interventionIntersection design, including number of turn lanes, width of medians, pedestrian overpass115 randomly selected signalized intersections with 4 legsPolice-reports of 585 bicycle-motor vehicle collisionsThree Pomodels oevent ris"throughvehicle trturn travright-turnsed only the term "accident" to describe crashes (collisions and/or falls) that may or may not havur reading of the studies, as explained in the "Safety terminology" section of the text. drives on the left (so turns should be interpreted accordingly), and bicycles travel on sidewalks ws that investigated relationships between bicyclist safety and intersection-related transpEnvironmental Health 2009, 8:47http://www.ehjournal.net/content/8/1/47Page 8 of 19(page number not for citation purposes)Table 3: Studies that investigated relationships between bicyclist safety and transportation infrastructure related to roads, lanes and/or paths.d Control method Effects observedate or es,  of ling Adjusted for distance traveled.Crash rates per million miles on major streets = 114, minor roads = 105, on-road bike routes or lanes = 58, and off-road = 292. Serious crash (involving emergency department visit or hospitalization) rates per million miles on major streets = 35, minor roads = 27, on-road bike routes or lanes = 25, and off-road = 77.ons  s, ral "Neutral" collision types (considered to be independent of bike lane presence) used as method to adjust for car-bike traffic on the different road types. Neutral collision types defined as those where the cyclist or motorist failed to stop or yield, or the motorist made an improper left turn.Bike lanes estimated to reduce collision frequency by 53%.Adjusted for average bicycle volumes city-wide in the before and after periods.Increase in crash rates with bike lanes, especially for lane on left side of street in the initial year post-intervention. No statistically significant effect on long-term crash rates.Reference Location; Design Infrastructure types examinedStudy population Outcome measuresAnalysis methoROADS, LANES AND PATHSKaplan(1975) Characteristics of the Regular Adult Bicycle User[61]United States;Observational, non-interventionMajor roads, minor roads, on-road bike routes or lanes, off-road (including bike paths and sidewalks)3,270 cyclists who completed a survey distributed to a random sample of League of American Wheelmen members, geographically weighted to represent the population of each state.Self-reporting (survey): 854 collisions or serious fallsCalculated crash rper million miles fdifferent infrastructure typbased on numbermiles cycled and proportion of cycon each type.Lott and Lott (1976)Effect of Bike Lanes on Ten Classes of Bicycle-Automobile crashes in Davis, California[62]United States - Davis; Observational, non-interventionRoads with and without marked bike lanes145 car-bike collisionsPolice reports of 145 car-bike collisionsComparison of numbers of collision roads with andwithout bike laneadjusting for neutcollision types.Smith and Walsh(1988)Safety impacts of bicycle lanes[63]United States - Madison;Observational, before-after interventionMajor roads with and without marked bike lanes (one on left side of street, one on right side)1.3-mile sections of 2 one-way arterial roadsCity-maintained database of traffic crashes*: 87 crashes at study sites (1,411 crashes city-wide)Compared crash counts per year before and after intervention./1/47Page 9 of 19(page number not for citation purposes)Tinsworth et al. United States; Major thoroughfares, (1) 420 cyclists who Hospital reports of Multiple logistic   .Adjusted for hours of bicycle use per month, age, sex, size of community, daylight vs. dawn/dusk/night.Relative risks (odds ratios) for injury by infrastructure type for adults: Major thoroughfares = 2.45; neighborhood streets (reference category) = 1; sidewalks = 1; bike paths = 0.14; unpaved surfaces = 0.11. Relative risks for children: neighborhood streets (reference category) = 1; sidewalks = 0.6; unpaved surfaces = 0.29; bike paths = 0.12.  Adjusted for miles traveled in warm weather months, age, sex, bicycle type, and geographic region of residence.Odds ratios for risk of being a cyclist who had collision or fall in the last year, by primary riding surface, compared to roadway (= 1.0): bike path or lane = 0.60; other surfaces = 1.28; off-road trail = 7.17.  Adjusted for distance traveled.Relative danger index by infrastructure type: major street without bike facilities = 0.66; minor street without bike facilities = 0.94; on-road bike routes = 0.51; on-road bike lanes = 0.41; multiuse trails = 1.39; off-road/unpaved trails = 4.49; "other" (mostly sidewalk) = 16.3.Table 3: Studies that investigated relationships between bicyclist safety and transportation infrastructure related to roads, lanes and/or paths. (Continued)Environmental Health 2009, 8:47http://www.ehjournal.net/content/8(1994) Bicycle-related injuries: Injury, Hazard, and Risk Patterns[64]Observational, non-interventionneighborhood streets, sidewalks, bike paths, unpaved surfaceswere injured and attended one of 90 emergency departments that report to the US Consumer Product Safety Commission, and (2) ~1250 other cyclists from a national probability sample420 injuries (emergency department visits)regression, comparinginfrastructure of injured cyclists (at location of injury event) and of cyclists from the national probability sample (infrastructure wherecyclist rode more than 50% of the time)Rodgers (1997)Factors Associated with the Crash Risk of Adult Cyclists[65]United States;Observational, non-interventionRoads, bike paths or lanes, off-road trails, other surfaces2,978 cyclists who completed a survey (conducted by National Family Opinion for Bicycling magazine), including adults who purchased new bicycles, screened to match US population based on geographic region, population density, household income, household size, and age.Self-reporting (survey): 280 respondents who had a crash or fell in the last 12 monthsMultiple logistic regression comparingodds ratios for havinga collision or fall versus not, according to primary riding surface of the cyclist.Moritz(1998)Adult Bicyclists in the United States: Characteristics and Riding Experience in 1996[66]United States;Observational, non-interventionMajor roads, minor roads, signed bike routes, on-street bike lanes, multiuse trails, off-road/unpaved trails, sidewalks1,956 cyclists who completed a survey distributed to a random sample of League of American Bicyclists members, geographically weighted to represent the population of each state.Self-reporting (survey): ~680 crashesRelative danger indices calculated by dividing the proportion of crasheson a given infrastructure type bythe proportion of commuting distance reported on that infrastructure. When index = 1.0, proportions of crashes and commuting distances are the same for that route type./1/47Page 10 of 19(page number not for citation purposes)nger ulated by   of crashes re type by tion of  distance n that re. When , s of d  distances e for that .Adjusted for distance traveled.Relative danger index by infrastructure type: major street without bike facilities = 1.26; minor street without bike facilities = 1.04; on-road bike routes and lanes = 0.50; off-road bike paths = 0.67; "other" (mostly sidewalk) = 5.3. calculated e traveled re type IS analyses   routes; ks for the structure ared using tribution  statistical Adjusted for distance traveled. Also adjusted (via weighting) for differences in use of various infrastructure types by cyclist characteristics: weekly commute distance; left turning method; comfort on busy streets; and belonging to a cycle club or having taken a training course.Compared to cycling on-road, there were no differences in collision rates for off-road or sidewalk cycling, but the relative risks of falls were 2.1 for off-road paths and 4.0 for sidewalks, and of injury were 1.6 for off-road paths and 4.0 for sidewalks. calculated e traveled re type IS analyses  commute ative risks ee re types using tribution  statistical Adjusted for distance traveled. Also adjusted (via weighting) for differences in use of various infrastructure types by cyclist characteristics: age; sex; weekly commute distance; and comfort on busy streets.Compared to cycling on-road, relative risks of collisions were 3.5 for off-road and 2.0 sidewalk cycling, of falls were 1.5 for off-road paths and 9.0 for sidewalks, and of injury were 1.8 for off-road paths and 6.4 for sidewalks.Table 3: Studies that investigated relationships between bicyclist safety and transportation infrastructure related to roads, lanes and/or paths. (Continued)Environmental Health 2009, 8:47http://www.ehjournal.net/content/8Moritz(1998)Survey of North American Bicycle Commuters: Design and Aggregate Results[67]United States;Observational, non-interventionMajor roads, minor roads, on-road bike routes & lanes, off-road bike paths, sidewalks2,374 cyclists who completed a survey distributed via email lists, magazine advertisements, and word of mouth.Self-reporting (survey): 271 serious crashesRelative daindices calcdividing theproportionon a given infrastructuthe proporcommutingreported oinfrastructuindex = 1.0proportioncrashes ancommutingare the samroute typeAultman-Hall and Hall (1998) Ottawa-Carleton commuter cyclist on- and off-road incident rates[68]Canada - Ottawa;Observational, non-interventionRoads, off-road paths, sidewalks1452 commuter cyclists who completed a survey distributed on parked bicycles.Self-reporting (survey): 187 injuries, 194 collisions, 234 fallsEvent ratesper distancon each infrastructubased on Gof mappedcommutingrelative risthree infratypes compPoisson disand Hauertest.Aultman-Hall and Kaltenecker (1999) Toronto bicycle commuter safety rates[29]Canada - Toronto;Observational, non-interventionRoads, off-road paths, sidewalks1196 commuter cyclists who completed a survey distributed on parked bicycles.Self-reporting (survey): 182 injuries, 300 collisions 203 fallsEvent ratesper distancon each infrastructubased on Gof mappedroutes; relfor the thrinfrastructucompared Poisson disand Hauertest.Environmental Health 2009, 8:47http://www.ehjournal.net/content/8/1/47Page 11 of 19(page number not for citation purposes)rdered  e 5 y Adjusted for traffic volume, speed limit, year, rural-urban, weather, daylight.More severe injuries were significantly associated with the following infrastructure characteristics: grades on both curved and straight roads; and unlit roads at night. Other factors associated with higher injury severity included: higher speed limits; lower average annual daily traffic; and fog.tic mparing els of y.Adjusted for whether child or adult, traffic volume per lane, household income, population density, land use, weather, and daylight.More severe injuries were significantly associated with wider roads, perceptible grades, and one-way streets, pavement not resurfaced in last 10 years, and highway road type (the first three variables at p < 0.05, the latter three at p < 0.10).d ogistic mparing severe  8) to s severe Adjusted for age, motor vehicle involvement, speed, helmet use.Decreased risk of severe injury on unpaved surfaces (odds ratio = 0.7, not statistically significant). Motor vehicle involvement was strongest risk factor (odds ratio = 4.6). lanes and/or paths. (Continued)ROAD DESIGN CHARACTERISTICSKlop and Khattak (1999)Factors Influencing Bicycle Crash Severity on Two-Lane, Undivided Roadways in North Carolina[69]United States - North Carolina;Observational, non-interventionCharacteristics of 2-lane undivided roads: curve vs. straight; level vs. grade; right shoulder width; intersection or not; street lighting1,025 collisions with motor vehicles.Police reports of bicycle collisions (recorded in the Highway Safety Information System) identifying injury severity†. Classified as property damage only, pain, non-incapacitating, incapacitating, and fatal.Multivariate oprobit modelcomparing thlevels of injurseverity.Allen-Munley et al.(2004) Logistic model for rating urban bicycle route safety[70]United States - Jersey City;Observational, non-interventionWidth and grade of roads, one-way versus two-way road configuration, highway versus non-highway road type314 injuries resulting from collisions with motor vehicles.Police reports of 314 bicycle crashes, identifying injury severity†. Classified as property damage only, minor and serious.Ordinal logisregression cothe three levinjury severitROAD SURFACESRivara et al. (1997) Epidemiology of bicycle injuries and risk factors for serious injury[16]United States - Seattle;Observational, non-interventionSurface type: paved vs. unpaved3390 injured cyclists who completed a questionnaire about demographic characteristics, cycling experience, crash circumstances, and helmet use and fit.Emergency department, hospital and medical examiner records of injuries, classified using the injury severity† score (ISS)Univariate anmultivariate lregression cocyclists with injuries (ISS >those with lesinjuries.Table 3: Studies that investigated relationships between bicyclist safety and transportation infrastructure related to roads,Environmental Health 2009, 8:47http://www.ehjournal.net/content/8/1/47Page 12 of 19(page number not for citation purposes) of r cyclists s vs. on isk  used 8-st counts ctions (7 d 2 with on the 3 Adjusted for age (whether child < 18 or adult), sex, and direction of travel (with or against motor vehicle traffic).Cycling on the sidewalk is associated with higher risk (RR = 1.8). The elevated risk on sidewalks is almost exclusively related to cycling against traffic (RR = 1.9) vs. with traffic (RR = 0.9). logitparing the of four ity Adjustment for all factors included in model: bicyclist age, intoxication, helmet use; driver intoxication; vehicle speed and type; crash characteristics including fault and directions of travel; land use; time of day; weather.Infrastructure-related determinants that increased the probability of severe injury in an crash were: unlit roads at night; curved road geometry; and undivided street configuration.estimating  in rsus lit versus Adjusted for hour of the day, darkness, and season, by summing log odds ratios calculated separately for these factors. Log odds ratios were weighted in inverse proportion to the variance of the odd ratio.Presence of lighting on rural roads reduces bicyclist injuries by ~60%.have substituted the words "injury", "crash", "collision" and/, lanes and/or paths. (Continued)SIDEWALKSWachtel and Lewiston (1994)Risk-factors for bicycle motor-vehicle collisions at intersections[71]United States - Palo Alto;Observational, non-interventionSidewalks vs. roadways89 bicycle-motor vehicle collisions at intersections or junctions on three major arterial roads.Police reports of 89 collisionsRelative riskcollisions foon sidewalkroadway. Rcalculationshour bicycliat 9 intersesignalized anstop signs) arterials.STREET LIGHTINGKim et al. (2007) Bicyclist injury severities in bicycle-motor vehicle crashes[72]United States - North Carolina;Observational, non-interventionStreet lighting, straight versus curved roadway, street configuration (one-way, two-way, divided or not)2934 injuries resulting from collisions between a single motorist and a bicyclist.Police reports of injury severity†. Classified as fatal; incapacitating; non-incapacitating; possible or no injury.Multinomialmodel, comprobability injury severoutcomesWanvik (2009)Effects of road lighting: an analysis based on Dutch crash statistics 1987-2006[73]The Netherlands;Observational, non-interventionRoad lighting on rural roads~125,000 bicycle crashes resulting in injury from 1987-2006.Police reports of ~125,000 injuriesOdds ratio risk of crashdarkness vedaylight on unlit roads.* This study used only the term "accident" to describe crashes (collisions and/or falls) that may or may not have resulted in injury. We or "fall" based on our reading of the studies, as explained in the "Safety terminology" section of the text.† Injury severity does not reflect risk of an incident, but rather the outcome of the incident once it occurs.Table 3: Studies that investigated relationships between bicyclist safety and transportation infrastructure related to roadsEnvironmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47given type of infrastructure was not studied in the English-language scientific literature identified by our search.MethodsSearch strategyWe searched the following bibliographic databases:Pubmed and Medline, which index over 3,600 interna-tional medical and health care journals (1949 to present);Web of Science, which includes the Science CitationIndex, the Arts and Humanities Citation Index, and theSocial Sciences Citation Index (1989 to present); andTransportation Research Information Services, whichincludes references to books, technical reports, conferenceproceedings and journal articles in the field of transporta-tion (1960 to present). In order to identify relevant stud-ies, we used search terms related to the safety of bicyclists,and to transportation infrastructure. Combinations of thefollowing keywords were used in the searches, (with"wildcards" used where appropriate to capture variants onterms, e.g. bicycl*): bicycle, safety, injury, accident, crash,conflict, infrastructure, road, and intersection. Referencelists of all relevant papers including review papers weresearched as a source of additional citations. The initial lit-erature search was conducted in summer 2008 andupdated through to June 2009.Inclusion and exclusion criteriaAll papers identified by the search were initially screenedfor relevance using the title and/or abstract. Specifically,we sought papers that met the description of injury epide-miology studies, injury severity studies, and crash/colli-sion/fall rate studies, and that considered some aspect ofinfrastructure as a determinant/predictor of bicyclists'safety. These included "before and after" studies thatexamine the safety impact (change in injury or crash ratefor cyclists) of some infrastructural change. Those papersconsidered potentially relevant were collected, and thefull text versions were then further reviewed for relevance.Papers were considered relevant and included in thereview if they met the following criteria:• they investigated a relationship between transportationinfrastructure (designed for either motorized or non-motorized use) and a clearly-defined metric of bicyclistsafety (injury, injury severity, crash/collision/fall); and• they were English-language publications describingempirical research conducted in an Organisation for Eco-nomic Co-operation and Development (OECD) country.For countries outside the OECD, it was expected that thetransportation infrastructure and bicycling rates (as wellas the socio-economic motivators of bicycling) would belocate any relevant papers describing studies conductedoutside the OECD.We excluded papers from further review if they met any ofthe following criteria:• studies of injuries or crashes that occurred when thebicycle was being used for bicycle racing, "off-road moun-tain-biking", trick/trials riding, or play;• studies only examining non-infrastructural determi-nants of safety such as helmet-use, bicycle type, personalcharacteristics of the bicyclists or motor vehicle drivers(e.g. age, sex, experience);• studies of injuries not related to a crash event, e.g.chronic injuries related to riding position;• studies examining gross numbers/types of injuries in aregion for a given time period, without either calculatingrates (per exposure/riding time) or considering infrastruc-tural determinants of those injuries;• studies that reported only subjective perceptions ofsafety or risk, whether by lay-public or experts; and• studies that examined only "conflict" between cyclistsand other road users (refer to the section on "safety termi-nology"), but where crashes or injuries were not identi-fied.ResultsIn total, 23 papers were identified that met the inclusioncriteria. Eight examined infrastructure related to intersec-tions, and are abstracted in detail in Table 2[53-60]. Fif-teen papers examined infrastructure related to"straightaways", i.e. roads, lanes, paths, etc., and areabstracted in Table 3[16,29,61-73]. Studies are presentedin the tables first by type of infrastructure, then by year foreach type.Ten of the 23 studies reviewed used injuries (or both inju-ries and crashes) as a metric of bicyclist safety, four exam-ined injury severity, and nine examined crashes (i.e.collisions and/or falls). Most of the studies were pub-lished since 1994, except two US studies which were pub-lished in the mid-70s [61,62] and one which waspublished in 1988 [63]. All the study designs were obser-vational. Five of the intersection-related papers [53,56-59], but only one of the road/lane/path-related papers[63], were "before-after" studies that quantified thechange in cyclist safety before and after some infrastruc-ture-related intervention took place. The remainingPage 13 of 19(page number not for citation purposes)different, and consequently the study results may not beapplicable across regions. The literature search did notpapers were classified as "non-intervention" observa-tional studies. Most of the studies based their analyses andEnvironmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47conclusions on at least 150 observations of injury or crashevents, and seven studies based their analyses on morethan one thousand observations. However one study ofroundabouts examined only 67 crashes, 58 of whichresulted in injuries [54], and two non-intersection studiesexamined 87 and 89 crashes on roads with and withoutmarked bike lanes [63], and on sidewalks versus roads[71] respectively.Thirteen of the studies were published in public healthrelated journals (mainly Accident Analysis & Preventionand the Journal of Safety Research), and nine were pub-lished in transportation engineering journals (mostlyTransportation Research Record). The remaining study(on the safety of different road/lane/path infrastructuretypes) was conducted as part of a Master of Science thesis[61].All but one of the studies about intersection-related infra-structure (Table 2) were conducted in European countries.Five of the European intersection-related studies exam-ined the safety of roundabouts and two examined markedbicycle crossings. The non-European study examined howintersection design in Japan influenced number of bicy-cle-motor vehicle collisions [60]. Cyclists in Japan arerequired by law to travel on the sidewalk, so the resultsfrom this study may not be generalizable to countries withdifferent traffic rules.The findings of the roundabout studies show some con-sistency, with elevated risks for cyclists after installation ofroundabouts with multiple traffic lanes or with markedbike lanes, whereas there were risk reductions or noapparent increase in risk at roundabouts with separatedcycle tracks [54,56,57]. One study showed a decreased riskfor cyclists and moped riders after installation of rounda-bouts in the Netherlands [53], but the authors did not dis-aggregate the results for these two road-user groups. Thefinding from this study - that roundabouts with separatedcycle tracks had a greater safety effect than those with on-road marked bike lanes or no bicycle infrastructure - isconsistent with other research. Another study on rounda-bout safety in Flanders found a similar effect for "vulner-able road users" [74], but we have not included this studyin our table because the vulnerable road user populationincluded pedestrians and motorized two-wheeler riders aswell as cyclists. It is likely that the safety effect of rounda-bouts, as measured in such "before-after" studies, willdepend on the "before" configuration of the intersectionsin question.The two studies of the safety effect of marked bicycle cross-ings at intersections looked at different design aspectsAlthough the study on elevated crossings showed a smallincrease in the number of crashes after the crossing wasinstalled, the bicycle traffic volume grew by 50% on thestreets after the intervention, as compared to unchangedstreets in the area, and this was not adjusted for in theanalysis [58]. The second study showed a reduction ininjury or crash risk when there was one colored bicyclecrossing at an intersection, but an increase in injury orcrash risk when there were two or more colored crossings[59].Of studies examining infrastructure related to straighta-ways on roads, lanes, and paths (Table 3), all but one wereconducted in Canada or the US. The only European studyin this category is very different in its focus: the safetyeffect of rural street lighting in the Netherlands [73]. Per-haps unsurprisingly, that study found that the presence ofstreet lighting on rural roads reduced the rate of cyclists'injuries by half. The effect was corroborated by an injuryseverity study that found that crashes resulting in moresevere injuries were significantly associated with unlitroads at night [69].Most of the remaining studies in this category comparedcyclist injury or crash rates on different types of road- orpath-related infrastructure that cyclists commonly travel,namely major and minor roads without specific cyclingfacilities, roads with wide curb lanes or marked bike lanes,on-road bike routes, off-road bike-specific or multi-usepaths, and sidewalks. A difficulty with this literature wasthat several facilities (between two and seven in number)were grouped into categories, such that facilities withpotentially different risks were classified within a singlecategory. In addition, the categorizations differed fromstudy to study, and the terminology used was sometimesnot clearly defined or consistently used. Despite these lim-itations, there are still some consistent messages.On-road marked bike lanes were found to have a positivesafety effect in five studies, consistently reducing injuryrate, collision frequency or crash rates by about 50% com-pared to unmodified roadways [61,62,65-67]. Three ofthose studies [61,66,67] found a similar effect for bikeroutes. One study [63] found that there was an increase incrash rates in the year following installation of markedbike lanes on a major road, especially for a section counterto on-road traffic flow, but this effect was not sustainedover the long term.There is less consistent evidence about off-road riding,possibly because this category encompassed a wide varietyof facility types. There may have been confounding factorssuch as whether the surface was paved or unpaved, or forPage 14 of 19(page number not for citation purposes)(one on physically elevated crossings, one on coloredcrossings) and did not provide clear conclusions.bicycles only or multiple user groups. Two studies exam-ined off-road bike paths and found reduced risks, rangingEnvironmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47from 0.11 to 0.67 times the risk of cycling on minor roads[64,67]. Two studies that grouped paved and unpaved,bicycle only and multi-use urban trails in their off-roadpath category found elevated risks, 1.6 to 3.5 times higherthan riding on-road [29,66,68]. Studies that examinedunpaved off-road trails as a separate category found risksof injury 2.5 to 7.2 times higher than on-road cycling[61,65,66] and 8 to 12 times higher than bike routes,lanes, or paths [65,66].Most studies that considered sidewalk-riding suggestedthat it is particularly hazardous for cyclists, with estimatesof 1.8 to 16 times the risk of cycling on-road [29,66-68,71]. However one study found that the risk of travelingon the sidewalk was the same or lower than riding on res-idential streets [64]. Another considered the direction oftravel and found that the elevated risk when sidewalkcyclists entered intersections was almost exclusivelyrelated to cycling against the flow of adjacent on-road traf-fic [71].Four studies examined the association between variousinfrastructural characteristics and injury severity[16,69,70,72]. More severe injuries were significantlyassociated with motor vehicle involvement, unlit roads atnight, wider roads, perceptible road grades, and one-waystreets. Injury severity does not reflect risk of an incident,but rather the outcome of the incident once it occurs. Incomparison, the studies that examined injury or crashrates, as opposed to those that concentrated on injuryseverity, were our primary focus since we are most inter-ested in shaping transportation infrastructure for injuryprevention.DiscussionIn this review we have described two categories of infra-structure: the first related to intersections; and the secondrelated to straightaways on roads, lanes, and paths. It is ofinterest to note that studies of the former type of infra-structure were conducted almost entirely in Europe, whilestudies of the latter were conducted almost entirely inNorth America. The reason for this may be the substantialdifferences in urban form, existing cycling infrastructure,cycling rates, and even the culture of cycling betweenEurope and North America. Pucher and colleagues havediscussed this issue extensively [26,75]. There is also sig-nificant variety in infrastructure design from one countryto another, and even within a given city. Despite this, ourreview has revealed that relatively few types of infrastruc-ture have been studied. For example, some common typesof infrastructure in North American cities have not beenassessed: traffic circles; bike boxes; sharrows; speedbumps/humps; and traffic diverters (Table 1). In addition,bicycle-specific design that is frequently available in highmodal share European cities. One of the limitations ofthis review is that we have only included studies in theEnglish scientific literature, although we are aware thatthere may be studies reported only in other (particularlyEuropean) languages.The principal trend that emerges from the papers reviewedhere is that clearly-marked, bike-specific facilities (i.e.cycle tracks at roundabouts, bike routes, bike lanes, andbike paths) were consistently shown to provide improvedsafety for cyclists compared to on-road cycling with trafficor off-road with pedestrians and other users. Marked bikelanes and bike routes were found to reduce injury or crashrates by about half compared to unmodified roadways.The finding that bicycle-specific design is importantapplies also to intersections with roundabouts, where itwas found that cycle tracks routing cyclists around anintersection separately from motor vehicles were muchsafer than bike lanes or cycling with traffic. It has beensuggested that the reason for high rates of bicycle-motorvehicle collisions at intersections is that motor vehicledrivers may be making "looked-but-failed-to-see" errors,whereby they search for oncoming motor vehicles but donot recognize that a cyclist is approaching because theyare not looking for them [39,40].Although roundabouts at intersections are not commonin North America, they are relatively popular in manyEuropean countries. It is possible that they may see morewidespread use in North America in the future because ofevidence that conversion of intersections to roundaboutsreduces crash risk for motor vehicle road users by 30-50%[76], especially when they replace intersections that werenot previously signal-controlled. However, because thecyclist-specific safety effect of roundabouts appears to behighly dependent on their design, transportation infra-structure planners should carefully consider interactionsbetween cyclists and other traffic modes. A literaturereview on the safety effect of roundabouts, prepared forthe 18th Workshop of the International Co-operation onTheories and Concepts in Traffic Safety [77], came to sim-ilar conclusions. It may be prudent to avoid installingroundabouts in areas where there is a high proportionalvolume of bicycle traffic, for example along designatedbicycle routes on residential roads. In some North Ameri-can cities there is retrofitting of "traffic circles" at intersec-tions in residential areas. Since these are quite differentfrom the larger-diameter roundabouts found in Europe,their effect on cyclist safety should be investigated beforemore widespread use is advocated.The reviewed literature also confirms some things thatPage 15 of 19(page number not for citation purposes)except for studies of roundabouts, we did not find anyinjury or crash studies that investigated cycle "tracks", amay already be "common-sense" for transportation plan-ners and safety experts: that streets used by cyclists at nightEnvironmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47should have good street-lighting, road surfaces should bepaved and well-maintained, and bike routes should avoidexcessive grades wherever possible.An issue with the literature to date, especially that relatedto roads, lanes, and paths, is that some investigators didnot define the terminology used. For example, the mean-ing of bike "path" was not defined in the paper by Tins-worth et al. [64]. Other investigators clearly defined theirinfrastructure terms, but grouped facilities that may havedifferent injury risks. For example, the studies of Aultman-Hall et al. [29,68] defined paths as "an off-road (usuallymulti-use) paved or unpaved path or trail," groupingpaths for bikes only, which were found by others to havelower risks than cycling on roads [64,67], with unpavedtrails, which were found by others to have higher risks[61,65,66]. Definitions of terminology are especiallyimportant in questionnaire-based studies to ensure thatstudy participants are all answering with the same infra-structure in mind; photos can be helpful in this regard[33].Clear and specific categorization is also vital to transpor-tation planners and engineers, so they can distinguishsometimes subtle differences between successful andproblematic design characteristics. One of the difficultiesof the studies in the English-language literature to date isthat the range of infrastructure studied is small comparedto the range of configurations used between and withinjurisdictions. Some examples are described above, butthere are many other features that merit investigation:stop signs; numbers of roads intersecting; junctions suchas driveways and lanes; cyclist lane of travel in relation toparked cars; surface features such as cobble stones orstreet-car (tram) tracks; traffic calming measures such asdiverters or road humps; and road/lane/path curvature.Underreporting of some events is an issue that is commonto all studies of bicycle injuries and crashes. Many of thestudies reviewed here relied on administrative datasources including hospital records [16,62,64], policereported accidents [54-61,69-73], and national or city-maintained registries [53,63], all of which are likely tomiss less severe events. For example, one of the large sur-veys [67] found that 9.8% of the respondents had had acrash in the last year, but only two in five crashes (38.2%)had been reported to police. Over half (56.6%) requiredmedical attention, but only one in twenty crashes (5.5%)required admission to a hospital. This underreportingmay create bias in infrastructure-specific risk calculations,since collisions involving motor vehicles may be morelikely to be reported to police for insurance reasons and tohospitals because they are more severe, as compared toof studies using these data sources should be interpretedas reflecting risk of severe events. Other studies in thisreview used data from cyclist surveys that may capture awider range of crash types, including those that are lesssevere [29,61,65-68]. However, survey data will not cap-ture events that resulted in fatalities (though these areextremely rare) or catastrophic incapacitating brain, spi-nal cord or other injuries and, depending on the methodof survey administration, may not capture individualswho no longer cycle following a crash [29,68]. No singlestudy design can overcome these reporting problems, thusthe importance of looking for consistency of results acrossdifferent designs.A great challenge in studying cycling injuries is ensuringthat comparisons control for the number of cyclists at risk(also called "exposure to risk"). The before-after studiesreviewed here aimed to do this by comparing numbers ofinjuries on the same intersection or roadway prior to andpost introduction of an infrastructure intervention, withthe assumptions that underlying traffic levels, injury rates,and types of cyclists stay the same. These assumptionsmay not hold [58], so some of these studies also adjustedfor temporal trends in traffic volumes [58,59,63] or injuryrates in the area [53], or made additional comparisons tounchanged intersections [56-59]. The non-interventionstudies needed to include methods to derive bicycling tripvolumes on the infrastructure types being compared.Sometimes these came from administrative data collectedby transportation authorities [54,55,60,71,73], and some-times from study participants describing the route of aninjury trip or their typical cycling location [29,61,64-68].Injury severity studies made comparisons within theinjured populations, so did not require trip volumedenominators [16,69,70,72], but this meant that theyexamined differences in severity of the outcome once inan injury event, not the original risk of the event itself.Though the most basic requirement for studies examiningrisk of crashes or injuries is to account for exposure to risk,there are many other factors that may confound compari-sons and that ideally would be controlled in study designor adjusted for in analyses. For example, men and womenor people in different age groups may choose to cycle ondifferent facility types, and might have different skill levelsor risk-taking behavior, thus creating the potential forconfounding associations between infrastructure andinjury. While it is difficult to control for all potential con-founders, many of the non-intervention studies reviewedhere did adjust for personal factors such as age[16,29,64,65,70,71], sex [29,64,65,71], cycling experi-ence [29,68], bicycle type [65], and environmental factorssuch as time of day [64,69,70,72,73] and weatherPage 16 of 19(page number not for citation purposes)collisions that happen with non-motorized users (whichmay happen more frequently on off-street paths). Results[65,69,70,72]. Most injury severity studies adjusted forhelmet use [16,69,72]. A style of observational study thatEnvironmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/47can control for most potential confounders is the case-crossover design [78]. Such a study is underway in theCanadian cities of Toronto and Vancouver. It will com-pare infrastructure at the injury site to that of randomlyselected control sites on the same trip, thus within-tripfactors (including age, sex, cycling experience, propensityfor risk taking, alcohol or drug use, bicycle type and con-dition, visibility via clothing or bicycle lights, weather,time of day, etc.) are controlled in the design.ConclusionAlthough the effect of infrastructure design on cyclistsafety was first studied more than three decades ago, theliterature on the topic remains remarkably sparse. Thisreview highlights opportunities for more detailed andcontrolled studies of infrastructure and cycling injuries.The evidence to date suggests that purpose-built bicycle-only facilities (e.g. bike routes, bike lanes, bike paths,cycle tracks at roundabouts) reduce the risk of crashes andinjuries compared to cycling on-road with traffic or off-road with pedestrians. Street lighting, paved surfaces, andlow-angled grades are additional factors that appear toimprove cyclist safety. The major advantage of infrastruc-ture modifications, compared to helmet use, is that theyprovide population-wide prevention of injury eventswithout requiring action by the users or repeated rein-forcement. Given the influence of safety on individuals'decisions to cycle, the importance of cycling modal shareto safety, and the ancillary benefits of this active and sus-tainable mode of transportation, infrastructure enhance-ments have the opportunity to promote an array ofimprovements to public health.AbbreviationsOECD: Organisation for Economic Cooperation andDevelopment; km: kilometer; ICD: International Classifi-cation of Diseases; AIS: Abbreviated Injury Scale.Competing interestsThe authors are part of a research team that is studying theassociation between bicyclists' injuries and the cyclingenvironment in Vancouver and Toronto. The Heart andStroke Foundation of Canada and the Canadian Institutesof Health Research have funded this three-year study. Seehttp://www.cher.ubc.ca/cyclingincities/injury.html formore information.Authors' contributionsKT, CR, AH and PC conceived of the study and developedthe literature search strategy. CR and AH conducted the lit-erature search. CR, AH, MW, and KT reviewed theincluded papers, and abstracted them for the detailedits development, particularly the discussion. All authorsreviewed and approved the final manuscript.AcknowledgementsThe authors thank Diana Kao who conducted an initial literature search that provided a base of material and search strategy for this review. We gratefully acknowledge the reviews by Jennifer Dill, Luc Int Panis, Russell Lopez and Anne Lusk, which helped improve the final paper. The authors would like to acknowledge the support of the University of British Colum-bia Bridge Program, the Heart and Stroke Foundation of Canada and the Canadian Institutes of Health Research. In addition, CR acknowledges fund-ing from the Transportation Association of Canada, and AH and MW acknowledge funding from the Michael Smith Foundation for Health Research.References1. Lindström M: Means of transportation to work and overweightand obesity: a population-based study in southern Sweden.Prev Med 2008, 46:22-28.2. Wen LM, Rissel C: Inverse associations between cycling towork, public transport, and overweight and obesity: findingsfrom a population based study in Australia.  Prev Med 2008,46:29-32.3. Gordon-Larsen P, Nelson MC, Beam K: Associations amongactive transportation, physical activity, and weight status inyoung adults.  Obes Res 2005, 13:868-875.4. Hamer M, Chida Y: Active commuting and cardiovascular risk:a meta-analytic review.  Prev Med 2008, 46:9-13.5. Cavill N, Davis A: Cycling and health: What's the evidence?Cycling England, UK Department of Transportation (Report) 2007.6. Gordon-Larsen P, Boone-Heinonen J, Sidney S, Sternfeld B, Jacobs DRJr, Lewis CE: Active commuting and cardiovascular diseaserisk.  Arch Intern Med 2009, 169:1216-1223.7. Woodcock J, Banister D, Edwards P, Prentice AM, Roberts I: Energyand health 3: energy and transport.  Lancet 2007,370:1078-1088.8. Boogaard H, Borgman F, Kamminga J, Hoek G: Exposure toultrafine and fine particles and noise during cycling and driv-ing in 11 Dutch cities.  Atmos Environ 2009, 43:4234-4242.9. Katzmarzyk PT, Janssen I: The economic costs associated withphysical inactivity and obesity in Canada: an update.  Can JAppl Physiol 2004, 29:90-115.10. Cavill N, Kahlmeier S, Rutter H, Racioppi F, Oja P: Economic anal-yses of transport infrastructure and policies including healtheffects related to cycling and walking: A systematic review.Transp Pol 2008, 15:291-304.11. Saelensminde K: Cost-benefit analyses of walking and cyclingtrack networks taking into account insecurity, health effectsand external costs of motorized traffic.  Transp Res A-Pol 2004,38(8):593-606.12. Thompson DC, Rivara FP, Thompson R: Helmets for preventinghead and facial injuries in bicyclists.  Cochrane Database Syst Rev.2000:CD001855.13. Cook A, Sheikh A: Trends in serious head injuries among Eng-lish cyclists and pedestrians.  Inj Prev 2003, 9:266-267.14. Linn S, Smith D, Sheps S: Epidemiology of bicycle injury, headinjury, and helmet use among children in British Columbia:a five year descriptive study.  Inj Prev 1998, 4:122-125.15. Macpherson AK, To TM, Macarthur C, Chipman ML, Wright JG, Par-kin PC: Impact of mandatory helmet legislation on bicycle-related head injuries in children: a population-based study.Pediatrics 2002, 110:e60.16. Rivara FP, Thompson DC, Thompson RS: Epidemiology of bicycleinjuries and risk factors for serious injury.  Inj Prev 1997,3:110-114.17. Thompson DC, Nunn ME, Thompson RS, Rivara FP: Effectivenessof bicycle safety helmets in preventing serious facial injury.JAMA 1996, 276:1974-1975.18. Chipman ML: Hats off (or not?) to helmet legislation.  CMAJ.Page 17 of 19(page number not for citation purposes)tables prepared for this paper. CR wrote the initial draft ofthe manuscript, and the other authors all contributed to2002, 166(5):602.19. Robinson DL: No clear evidence from countries that haveenforced the wearing of helmets.  BMJ 2006, 332:722-725.Environmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/4720. Nelson AC, Allen D: If you build them, commuters will usethem: association between bicycle facilities and bicycle com-muting.  Transp Res Rec 1997, 1578:79-83.21. Lopez RP, Hynes HP: Obesity, physical activity, and the urbanenvironment: public health research needs.  Environ Health2006, 5:25.22. Dill J, Carr T: Bicycle commuting and facilities in major US cit-ies: If you build them, commuters will use them.  Transp ResRec 2003, 1828:116-123.23. Cervero R, Sarmiento O, Jacoby E, Gomez L, Neiman A: Influencesof built environments on walking and cycling: Lessons fromBogota.  Int J Sust Transp 2009, 3:203-226.24. Moudon AV, Lee C, Cheadle AD, Collier CW, Johnson D, Schmid TL,Weather RD: Cycling and the built environment, a US per-spective.  Transp Res D-Transp Environ 2005, 10:245-261.25. Frumkin H, Frank LD, Jackson RB: Urban sprawl and public health:Designing, planning, and building for healthy communities Washington,DC: Island Press; 2004. 26. Pucher J, Buehler R: Making cycling irresistible: lessons fromthe Netherlands, Denmark and Germany.  Transp Rev 2008,28:495-528.27. Pucher J, Dijkstra L: Promoting safe walking and cycling toimprove public health: lessons from the Netherlands andGermany.  Am J Public Health 2003, 93:1509-1516.28. Pucher J, Buehler R: Why Canadians cycle more than Ameri-cans: a comparative analysis of bicycling trends and policies.Transp Pol 2006, 13:265-279.29. Aultman-Hall L, Kaltenecker MG: Toronto bicycle commutersafety rates.  Accid Anal Prev 1999, 31:675-686.30. Ogilvie D, Egan M, Hamilton V, Petticrew M: Promoting walkingand cycling as an alternative to using cars: systematic review.BMJ 2004, 329:763.31. Carver A, Salmon J, Campbell K, Baur L, Garnett S, Crawford D:How do perceptions of local neighborhood relate to adoles-cents' walking and cycling?  Am J Health Promot 2005, 20:139-147.32. Noland R: Perceived risk and modal choice: risk compensa-tion in transportation systems.  Accid Anal Prev 1995, 27:503-521.33. Winters M, Teschke K: Route preferences among adults in thenear market for bicycling: findings of the Cycling in Citiesstudy.  Am J Health Promot; 2009 in press.34. Decima Research Inc: City of Toronto 1999 cycling study: finalreport on quantitative research results.  (Report) 2000 [http://www.toronto.ca/cycling/reports/pdf/decimareport.pdf ].35. Robinson DL: Safety in numbers in Australia: more walkersand bicyclists, safer walking and bicycling.  Health Promot J Austr2005, 16:47-51.36. Jacobsen P: Safety in numbers: more walkers and bicyclists,safer walking and bicycling.  Inj Prev 2003, 9:205-209.37. Vandenbulcke G, Thomas I, de Geus B, Degraeuwe B, Torfs R,Meeusen R, Int Panis L: Mapping bicycle use and the risk of acci-dents for commuters who cycle to work in Belgium.  TranspPol 2009, 16:77-87.38. Elvik R: The non-linearity of risk and the promotion of envi-ronmentally sustainable transport.  Accid Anal Prev 2009,41:849-855.39. Herslund M, Jorgensen N: Looked-but-failed-to-see-errors intraffic.  Accid Anal Prev 2003, 35:885-891.40. Rasanen M, Summala H: Attention and expectation problems inbicycle-car collisions: an in-depth study.  Accid Anal Prev 1998,30:657-666.41. CDC: Motor-vehicle safety: a 20th century public healthachievement.  JAMA 1999, 281:2080-2082.42. International Statistical Classification of Diseases andRelated Health Problems, 10th Revision (ICD-10)   [http://apps.who.int/classifications/apps/icd/icd10online/]43. Carter D, Hunter WW, Zegeer C, Stewart JR, Huang H: Bicyclistintersection safety index.  Transp Res Rec 2007,2031(2007):18-24.44. Hydén C, Várhelyi A: The effects on safety, time consumptionand environment of large scale use of roundabouts in anurban area: a case study.  Accid Anal Prev 2000, 32:11-23.45. Hunter WW: Evaluation of innovative bike-box application inEugene, Oregon.  Transp Res Rec 2000, 1705(2000):99-106.46. Alta Planning + Design: San Francisco's Shared Lane Pavementwww.sfmta.com/cms/uploadedfiles/dpt/bike/Bike_Plan/Shared%20Lane%20Marking%20Full%20Report-052404.pdf].47. Hallett I, Luskin D, Machemehl R: Evaluation of On-Street Bicy-cle Facilities Added to Existing Roadways.  (Report) FHWA/TXDOT-06/0-5157-1 2006 [http://www.utexas.edu/research/ctr/pdf_reports/0_5157_1.pdf].48. Hunter WW, Feaganes JR, Srinivasan R: Conversions of wide curblanes: the effect on bicycle and motor vehicle interactions.Transp Res Rec 2005, 1939(2005):37-44.49. Hunter WW, Stewart JR, Stutts JC: Study of Bicycle Lanes VersusWide Curb Lanes.  Transp Res Rec 1999, 1674(1999):70-77.50. Jordan G, Leso L: Power of the line - Shared-use path conflictreduction.  Transp Res Rec 2000, 1705(2000):16-19.51. Neira J, Bosque L: The word "accident": no chance, no error,no destiny.  Prehosp Disaster Med 2004, 19(3):188-189.52. Davis RM, Pless B: BMJ bans "accidents".  BMJ 2001,322:1320-1321.53. Schoon C, Van Minnen J: The safety of roundabouts in the Neth-erlands.  Traffic Eng Control 1994, 35:142-148.54. Brüde U, Larsson J: What roundabout design provides the high-est possible safety?  Nordic Road Transp Res 2000, 2:17-21 [http://www.nordicroads.com/website/files/Nordic_nr2-2000.pdf].55. Hels T, Orozova-Bekkevold I: The effect of roundabout designfeatures on cyclist accident rate.  Accid Anal Prev 2007,39:300-307.56. Daniels S, Nuyts E, Wets G: The effects of roundabouts on traf-fic safety for bicyclists: an observational study.  Accid Anal Prev2008, 40:518-526.57. Daniels S, Brijs T, Nuyts E, Wets G: Injury crashes with bicyclistsat roundabouts: influence of some location characteristicsand the design of cycle facilities.  J Safety Res 2009, 40:141-148.58. Gårder P, Leden L, Pulkkinen U: Measuring the safety effect ofraised bicycle crossings using a new research methodology.Transp Res Rec 1998, 1636(1998):64-70.59. Jensen SU: Safety effects of blue cycle crossings: a before-afterstudy.  Accid Anal Prev 2008, 40:742-750.60. Wang Y, Nihan NL: Estimating the risk of collisions betweenbicycles and motor vehicles at signalized intersections.  AccidAnal Prev 2004, 36:313-321.61. Kaplan J: Characteristics of the regular adult bicycle user.  InMSc thesis University of Maryland, Civil Engineering Department;1975. 62. Lott DF, Lott DY: Effect of Bike Lanes on Ten Classes of Bicy-cle-Automobile Accidents in Davis, California.  J Safety Res1976, 8:171-179.63. Smith R Jr, Walsh T: Safety impacts of bicycle lanes.  Transp ResRec 1988, 1168:49-56 [http://pubsindex.trb.org/view.aspx?id=295848].64. Tinsworth D, Cassidy S, Polen C: Bicycle-related injuries: Injury,hazard, and risk patterns.  Int J Inj Contr Saf Promot 1994,1(4):207-220.65. Rodgers GB: Factors associated with the crash risk of adultbicyclists.  J Safety Res 1997, 28(4):233-241.66. Moritz WE: Adult bicyclists in the United States: characteris-tics and riding experience in 1996.  Transp Res Rec 1998,1636(1998):1-7.67. Moritz WE: Survey of North American bicycle commuters:design and aggregate results.  Transp Res Rec 1998,1578(1998):91-101.68. Aultman-Hall L, Hall FL: Ottawa-Carleton commuter cyclist on-and off-road incident rates.  Accid Anal Prev 1998, 30:29-43.69. Klop JR, Khattak AJ: Factors influencing bicycle crash severityon two-lane, undivided roadways in North Carolina.  TranspRes Rec 1999, 1674:78-85.70. Allen-Munley C, Daniel J, Dhar S: Logistic model for rating urbanbicycle route safety.  Transp Res Rec 2004, 1878:107-115.71. Wachtel A, Lewiston D: Risk-factors for bicycle motor-vehiclecollisions at intersections.  Inst Transp Eng J 1994, 64:30-35.72. Kim JK, Kim S, Ulfarsson GF, Porrello LA: Bicyclist injury severi-ties in bicycle-motor vehicle accidents.  Accid Anal Prev 2007,39:238-251.73. Wanvik PO: Effects of road lighting: an analysis based onDutch accident statistics 1987-2006.  Accid Anal Prev 2009,41:123-128.Page 18 of 19(page number not for citation purposes)Markings: Improving Bicycle Safety.  (Report) 2004 [http://Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central Environmental Health 2009, 8:47 http://www.ehjournal.net/content/8/1/4774. De Brabander B, Vereeck L: Safety effects of roundabouts inFlanders: signal type, speed limits and vulnerable road users.Accid Anal Prev 2007, 39:591-599.75. Pucher J, Komanoff C, Schimek P: Bicycling renaissance in NorthAmerica? Recent trends and alternative policies to promotebicycling.  Transp Res A-Pol 1999, 33(7-8):625-654.76. Elvik R: Effects on road safety of converting intersections toroundabouts: review of evidence from non-US studies.Transp Res Rec 2003, 1847:1-10.77. Daniels S, Wets G: Traffic safety effects of roundabouts: areview with emphasis on bicyclist's safety.  Proceedings of the18th ICTCT Workshop; 27-28 October 2005; Helsinki, Finland 2005[http://www.ictct.org/dlObject.php?document_nr=25&/S3_Daniels.pdf ].78. Maclure M: The case-crossover design: a method for studyingtransient effects on the risk of acute events.  Am J Epidemiol1991, 133:144-153.yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 19 of 19(page number not for citation purposes)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items