Content uploaded by Chris Rissel
Author content
All content in this area was uploaded by Chris Rissel on Oct 14, 2014
Content may be subject to copyright.
1
Exposure‐basedcyclingcrash,nearmissandinjuryrates:TheSaferCyclingProspectiveCohort
Studyprotocol
Authors
RoslynG.Poulos1
JulieHatfield2,
ChrisRissel3
RaphaelGrzebieta2
AndrewSMcIntosh4
Affiliationsandcontactdetails
1SchoolofPublicHealthandCommunityMedicine,TheUniversityofNewSouthWales,Sydney
2052,Australia.
2TransportandRoadSafety(TARS)Research,TheUniversityofNewSouthWales,Sydney,New
SouthWales,Australia.
3SchoolofPublicHealth,SydneyMedicalSchool,UniversityofSydney,Sydney,Australia.
4RiskandSafetyScience,TheUniversityofNewSouthWales,Sydney,NewSouthWales,Australia.
Correspondenceto
RoslynGPoulos,SchoolofPublicHealthandCommunityMedicine,TheUniversityofNewSouth
Wales,Sydney2052,Australia;r.poulos@unsw.edu.au
2
Exposure‐basedcyclingcrash,nearmissandinjuryrates:TheSaferCyclingProspectiveCohort
Studyprotocol
Abstract
Introduction:Thereareclearpersonal,socialandenvironmentalbenefitsofcycling.However,
safetyconcernsareamongthefrequentlycitedbarrierstocycling.InAustralia,thereareno
exposure‐basedmeasuresoftheratesofcrashor‘nearmiss’experiencedbycyclists.
Designandsetting:Aprospectivecohortstudyover12months,withalldatacollectedviaweb‐
basedonlinedataentry.
Participants:Twothousandadultsaged18yearsandolder,livinginNewSouthWales(Australia),
whousuallybicycleatleastonceamonth,willberecruitedfromMarchtoNovember2011.
Methods:Inthe12monthsfollowingenrolment,cyclistswillbesurveyedon6occasions(weeks8,
16,24,32,40,and48fromtheweekoftheenrolmentsurvey).Inthesesurveyweeks,cyclistswillbe
askedtoprovidedailyreportsofdistancetravelled;time,locationanddurationoftrips;
infrastructureused;crashes,nearmissesandcrash‐relatedinjuries.Informationoncrashesand
injurieswillalsobesoughtfortheinterveningperiodbetweenthelastandcurrentsurvey.A
subsampleofparticipantswillreceivebicycletripcomputerstoprovideobjectivemeasurementof
distancetravelled.
Discussion:Thisstudyprotocoldescribestheprospectivecohortstudydevelopedtoassessnear
misses,crashesandinjuriesamongcyclistsbytimeanddistancetravelledandbytypeof
infrastructureused,withrecruitedparticipantsenteringdataremotelyusingtheinternet.Weexpect
tobeabletocalculateeventrateaccordingtoexposureoverallandfordifferentinfrastructuretypes
andtoreportin‐depthinformationabouteventcausation.
3
Exposure‐basedcyclingcrash,nearmissandinjuryrates:TheSaferCyclingProspectiveCohort
Studyprotocol
Background
Thereareclearpersonal,socialandenvironmentalbenefitsofcycling.Thesebenefitsincreaseas
individualscyclemoreandasmorepeoplecycle.[1]Personalbenefitsofrecreationalortransport
cyclingincludesubstantialhealthgains,[2]withasignificantreductioninall‐causemortalityand
cardiovascularrisk.[3,4]Cyclingmayalsobeanimportanttoolinthefightagainsttheobesity
epidemic,withecologicalstudiesnotinglowerratesofobesityinregionswithhighratesof
cycling.[5]ThevalueofcurrentcyclingtotheAustralianhealthsystemhasbeenconservatively
estimatedat$227millionperannum.[1]
Cycling,andurbanenvironmentsthatsupportcyclingandwalkinganddiscouragecaruse,can
improvesocialinteractionsandincreasecommunityattachment,liveabilityandamenity.[6]Thereis
alsoevidencethatthemorecompact,permeableurbandesignsthatsupportcyclingandwalking
leadtocrimereductionthroughincreasedstreetactivityand‘naturalsurveillance’.[7]
Wherecyclingreplacesamotorvehicletripthereareenvironmentalbenefits,suchasreduced
carbondioxideemissionsandotherpollutants.[1]Additionally,anincreaseintransportcyclingmay
easetrafficcongestion,reducemotorvehiclecrashesandthuspotentiallyreduceroadtrauma
costs.[8]
ThetransportationcyclingrateinAustraliaislowcomparedtootherdevelopedcountriessuchas
theNetherlands,DenmarkorGermany.[5]Countrieswithhighcyclingratestendtohavebetter
cyclingfacilities,effectivetrafficcalmingmeasuresandwell‐developedcyclinginfrastructure,
cycling‐friendlyurbandesign,restrictionsonmotorvehicleuseandcomprehensivetrafficeducation
abouthowtointeractwithcyclists.[9]
Cycling‐specificinfrastructureisgenerallyrecognisedasanecessarystartingpointtoattractnew
non‐cycliststocycling.[1]However,itisnotalwaysclearwhattypeofinfrastructurewillproducethe
maximaladoptionofcyclingwiththegreatestsafetybenefits.Forexample,introducingshared‐use
pathsforpedestriansandcyclistsmaydecreasecyclistcrashrateswithmotoristsbutcouldincrease
pedestrianinjuryrates.[10]
4
SignificantnumbersofcyclistsarekilledorseriouslyinjuredinAustraliaeveryyear,andthese
numbersappeartobeincreasing.Morethan15000bicyclistswereinvolvedinpolicereported
crashesintheAustralianstatesofVictoria,Queensland,SouthAustralia,andWesternAustralia
during2000‐2004,[11]andthereweremorethan2000motor‐vehicle‐relatedcrashesleadingto
hospitalisationsinbicyclistsinthestateofNewSouthWalesbetween1999/2000and2004/2005.
[12]Further,itislikelythatthesedataunderenumeratetheproblembecausemanylessserious
cyclingincidentsgounreportedtopolice,orbecausethedatadonotincludedeathsandinjuries
arisingfromfallsandcollisionsnotinvolvingmotorvehiclesanddonotincludecasualtiesreceiving
outpatienttreatment.RecentlypublisheddatafromVictoriareportsasignificantincreasein
emergencydepartmentandhospitaladmissionsforbicycleinjuriesfrom2001to2006andamarked
increaseinthenumberofcyclistssustainingsevereinjury.[13]However,theinterpretationofthese
dataisseverelylimitedbythequalityofdataintermsofcyclingexposureandcontributing
factors.[13,14]
Toreducetherisksassociatedwithcyclistssharingtrafficlaneswithmotorvehicles,strategiessuch
asbicyclelanes(i.e.,portionsoftheroadwaydesignedfortheuseofbicycles)andbicyclepaths(a
physicallyseparatepathwayforbicyclesorashared‐usepathforbicyclesandpedestrians)have
beenbuilt.Therelativemeritsofthesedifferentstrategiesremaindebated,especiallyinurbanareas
inmajorcities,wherethisquestionisimportantduetolimitedavailabilityoflandandfinance.
Researchduringthe1970sintheUSAindicatedthatbicyclelanesmayhaveaprotectiveeffectfor
mostcollisiontypes,butanincreasedriskforcollisionsoccurringwhenthecyclistturnsleftinto
traffic(turningrightinAustralia).[15]Laterresearchfoundthattheriskofridingonthesidewalk
(includingbicyclepathsandfootpaths)washigherthantheriskofridingontheroadway,probably
becauseofblindconflictsatintersections.[16]EarlyresearchfromEnglandreportedthattherateof
injurywasgreateroncyclewayscomparedtolocalroadsandgridroads,butprimarilybecauseof
poordesignincludingvisibilityproblems,especiallyatintersections,sharpbends,steepgradients,
slipperybridges,loosegravelandmud.[17]Otherrisksincludedhead‐oncrashesbetweencyclists,
collisionswithdogs,andeyeinjuriesfromintrudingvegetation.[17]Thecontestednatureofthe
evidencesupportingvariousinfrastructuresolutionshasledforcallsintheinternationalliterature
formoreresearcharoundinfrastructureandforimprovedunderstandingaboutbehaviour,safety
andsafetyperception.[18]
Ifcyclingistobeencouragedasahealth‐promotingpractice,orasasustainableformoftransport,
thenitisincumbentonauthoritiestoprovideanenvironmentthatminimisesrisk,tomanagepublic
5
perceptionofriskandtopromotetheactivityappropriately.However,becauseofthelackof
exposuredatainAustralia,[13]actualcrashandinjuryratesrelatedtodistancecycled,durationof
travelorinfrastructureusedareunknown.
CyclingisbeingactivelypromotedbytheAustralianhealthandtransportsectorsandbygovernment
atalllevels,[19‐21]andinfrastructureisbeingprogressivelydevelopedacrossthecountry.
However,thereisverylittleknowledgeonwhichtobasedecisionmaking.Thisprojectistherefore
significantbecauseitmeetsanurgentandrapidlyincreasingdemandforevidencetosupportpolicy.
Thisstudyaimsto:
1. Developmeasuresofcyclistcrash,nearmissandinjuryratesforabroadlyrepresentative
populationofcyclistsfrommetropolitanandregionalNewSouthWales
2. Identifyfactorsthatcontributetocrash,nearmissandinjuryrates,forexample,human
factorsandroadenvironment.
3. Assesstherisksforcyclistsassociatedwithcyclingonroads,bicyclelanesandpaths.
4. Describethetype,locationandfrequencyofhazardsidentifiedbycyclists.
Methods
Overallstudydesign:Acohortstudyoftwothousandcyclists(18yrsandover)whoresideinNew
SouthWalesandrideatleastoncepermonthwillbeundertaken.Cyclistswillberecruitedvia
multiplechannelsincludingtheextensiveemaillistsofBicycleNSW(astatecyclingadvocacy
organisation),arangeofcommunitycyclingevents,bikeshops,mediapublicityandthroughwordof
mouthwithinthecyclingcommunity.Participantswillenrolandcompletethebaseline
questionnaireandallsubsequentfollow‐upquestionnairesviaasecurewebsite.Automatically
generatedemailsplusSMStextmessages(forthosewhoselecttextreminders)willbesentto
participantswhenfollow‐upquestionnairesaredueforcompletion.
Toincreasethelikelihoodofaccuratemeasurementofdistancesandtimetravelled,bicycletrip
computerswillbeofferedtoasubsetofenrolledcyclistswhorideatleastonceperweekandwho
reportnothavingtheirowntripcomputersonenrolment.Toensurethemosteffectiveuseof
limitedresources,cyclistswillnotbeofferedtripcomputersiftheyhavemorethanonebikeanddo
notidentifyridingonebikeatleast90%ofthetime,iftheirmainbikeisaBMXbikeas‘offroad’
cyclingisoutsidetheprimaryaimofthestudyoriftheirmainbikeisclassifiedas‘other’andmay
6
notbecompatiblewiththebicycletripcomputer(e.g.,folding,recumbentorelectricbikes).
Participantswishingtoaccepttheofferofatripcomputerwillbeaskedtosupplythedataontyre
sizefromthewalloftheirbicycletyreandthediameteroftheirwheel.Tripcomputersare
individuallyprogrammedbyresearchstaffusingthetyrecircumferencereferencetableor
calculationinstructionssuppliedbythemanufacturer.Tripcomputerswiththemanufacturer’s
installationinstructionswillbesentbyposttoparticipants.Bicycletripcomputerswillbeoffered
untilresearchstocksareexhausted(estimatedtobearound400computers).
Pilottesting:Paperversionsofthequestionnaireswerepilotedwithcolleagueswhocycle(within
theUniversityandfundingpartnerorganisations)andmodifiedaccordingly.Electronicversions
wereagaintestedwithcolleagueswhocycle,priortoimplementation.
Baselinedatacollection:Atenrolment,demographicdata(e.g.,age,gender,education,
employmentstatus,income,drivinglicence,caraccess),cyclingexperienceandconfidence,self‐
identificationofcyclisttype(transportorrecreational),usualcyclinghabits(averagemonthly
distanceandhourscycled,useofinfrastructureandcyclingequipmentoverpast12months),history
ofcrashesandcrash‐relatedinjuriesoverthelast12months,andattitudestoriskandsensation
seeking(usingpreviouslyvalidatedinstruments[22,23])willbecollected.
Follow‐updatacollection:Overthe12monthsfollowingenrolment,therewillbe6surveyweeks
(weeks8,16,24,32,40,and48fromtheweekofthebaselinesurvey).Atthestartofeachsurvey
week,participantswillbeaskedaboutcrashesandcrash‐relatedinjuriesfortheinterveningperiod
betweenthelastandcurrentsurvey,andthentocompleteatraveldiaryfortheensuing7days.
Cyclistswillbeaskedtoprovidedailyreportsofdistancetravelled;time,locationanddurationof
trips;infrastructureused;crashes,nearmissesandcrash‐relatedinjuriesexperienced.Toenablethe
calculationofrates,participantswithbiketripcomputerswillbeaskedtoprovidereadingsfromthe
computer,whileotherparticipantswillestimatetimeanddistancestravelled.Cyclistsmayenter
dataviathesecurewebsiteonadailybasis,orkeepahardcopyrecordoftheirtripsbyprinting
downa7‐dayPDFversionofthediary,andenteringtheirdataattheendoftheweek.
Ateachsurvey,severaladditionalquestionsonaspectsofcyclingwillbeasked.Thesewillinclude
aggressionexperiencedwhilecyclingwithotherroadusers(follow‐upperiod1);infrastructure
preferencesandconcerningsafetyissues(follow‐upperiod2);cyclingactivitiesandbeliefsinrespect
7
ofsafety(follow‐upperiod3);observationoftheroadrulesandlawsforcyclists(follow‐upperiod
4);cyclingfortransport(follow‐upperiod5);andcyclingwithchildren(follow‐upperiod6).
Duringeachfollow‐upperiod,participantswillreceivereminderemails(oneonthedaypriortothe
commencementofthereportingperiod,ondays2to7andafinalreminderemailonday8).The
computerprogrammingallowsfor7consecutivedaysofreportingwithina14‐daywindow,after
whichaccesstothefollow‐upquestionnaireisblocked.Thisflexibilitywillallowcycliststhe
opportunityofcommencingtheirdiaryupto7daysafterthestartdateandstillcollecttheirdata
prospectively(e.g.,incaseswherecyclistsdonotaccesstheiremaildailyandthereforereceivelate
notificationoftheirstartdate)ortoallowsomeflexibilityinthenumberofdaysavailableinwhich
datacanbeenteredforthosecyclistswhohavekeptahardcopydiary.
Qualitativedatacollectionfollowingcrashreports:Allcyclistsreportingacrashoccurringduring
theirreportingweekwillbecontactedbytelephonetoansweradditionalquestionsaboutthe
locationoftheevent,thecircumstancesleadingtotheevent,behaviouralorenvironmentalrisk
factors,injuriessustained(type,severityandtreatment),andsuggestedwaysinwhichtheevent
couldhavebeenprevented.
Definitions:Crashesaredefinedascollisionsorfalls,basedonthedefinitionsgiveninthereviewby
Reynoldsetal(2009).[24]Acollisionisdefinedasaneventinwhichthebicyclehitsorishitbyan
object,personoranimal,regardlessoffault;andafallisdefinedasanevent(notcausedbya
collision)wherethebicycleand/orbikeriderlandsontheground.
Anearmissisdefinedasanunexpectedeventwhilecyclingthatcausesthecyclistoranotherparty
totakesuddenevasiveaction,andwithoutsuchactionthecyclistbelievesacrash(collisionorfall)
wouldhavehappened.
Forclarity,participantswillbeprovidedwithimagesanddefinitionsofallrelevantcycling
infrastructure,asshowninfigure1.
8
Figure1Cyclinginfrastructuredefinitions.
Planneddataanalysis:Quantitativedataanalysiswillincludebasicdescriptiveanalysisofthe
baselineandfollow‐updataandacalculationofeventrates.Eventrateswillbecalculatedusingthe
incidentnumberofevents(collision,fallsornearmisses)perhoursofcyclingorkilometrestravelled.
Theincidencerateofcrasheswillalsobecalculatedfordifferentformsofinfrastructure.
Multivariablegeneralisedlinearmodellingwillbeusedtoexplorehumanfactors(cyclist
characteristicssuchascyclingexperience,attitudesandbehaviour)aspredictorsofoutcome(event
rate).
Qualitativedatawillbeanalysedusing‘templateanalysis’.[21]Thismethodinvolvesthe
developmentofacodingtemplateorframeworkcomposedofcodesrepresentingthemesidentified
inthedatathroughmultiplereadingsofthetext.
Discussion
Thisstudyprotocoldescribesanoriginalresearchplanthatwillenablethecalculationofratesof
crashes,nearmissesandinjuriesrelatingtotimeanddistancecycledandthetypeofinfrastructure
usedforcycling.Thedeterminationofexposure‐basedrateswillmakeasignificantcontributionto
knowledgeaboutcyclinginAustralia,astheonlymeasurescurrentlyavailablearebasedon‘per
head’ofpopulation.[13].Itwillalsoallowcomparisonsofcyclingrisktoothertravelmodesin
Australiaandinternationally.Additionalquestionsaboutcyclingexperience,attitudesand
behaviourwillenabletheexplorationofthesevariablesaspredictorsofoutcome.Thestudywillalso
helptoanswerquestionswhicharedebatednationallyandinternationallyaroundtherelativemerit
ofdifferentformsofinfrastructure.Inaddition,wewillrecordcrashesandinjuriesthatarenot
capturedintheavailabledatacollections,suchascrasheswhicharenotreportedtopoliceor
injurieswhichdonotrequirehospitalisation.Capturingqualitativeandquantitativedataonthese
9
crasheswillprovideagreaterunderstandingofthesurroundingcircumstancesandislikelytomeet
theneedforadditionalinformationonsingle‐vehiclebicyclecrashes.[25]Itwillenableustobetter
understandcyclingriskandtoprovideevidencetosupportabest‐practicesystemofcycling
infrastructureandeffectivepubliceducation.Thisshouldhelptopromotebettermanagementof
interactionsbetweenallusersoftransportfacilitiesinthefuture.
Strengthsandlimitations
Thisstudyisunique.Toourknowledge,itisthefirsttomeasureexposure‐basedratesofcrashes,
nearmissesandcrash‐relatedinjuriesforcyclistsinAustralia.Itwillbeconductedcompletely
online,supportedbyavarietyofonlineengagementandinteractionactivitiessuchastheSafer
CyclingStudyFacebookpage,aYouTubeclipaboutthestudyandregularemailand/ortextmessage
phonereminderstostudyparticipants.Thesearedesignedtomaintaincohortparticipationthrough
alldatacollectionstages.Inaddition,allcrashesrecordedindailyreportswillbefollowedupby
interviewtoexploreaspectsofthecrashthatarenotelicitedbythequantitativequestionsandto
provideacontextualunderstandingofthecrashevent.Finally,theprospectivenatureofthestudy
isexpectedtoreducerecallbias,andtheuseofbicycletripcomputersshouldreducemeasurement
error.
Thestudyislikelytohaveanumberoflimitations.Theseincludethepotentialforvolunteerandself‐
reportbias.Losstofollow‐upisalsolikely.However,asparticipantshavebeenaskedtosupplya
numberofdifferentcontactoptions(includingthoseofafriendorfamilymember),weexpecttobe
abletoidentifyparticipantsinwhomlosstofollow‐uphasbeentheresultofaseriouscycling‐
relatedinjury.
Acknowledgements
WethankDrSusanneMurphyforherassistanceinpreparingthismanuscriptforpublicationandthe
representativesfromourfundingpartnerorganisationswhohavecontributedtodevelopmentand
implementationofthisstudy.
Funding
ThisprojecthasbeenfundedundertheAustralianResearchCouncil’sLinkageProjectsfunding
scheme(projectnumberLP1000100597)withfinancialcontributionsfromtheRoadsandTraffic
10
AuthorityofNewSouthWales,SydneySouthWestAreaHealthService,BicycleNSWandWilloughby
Council.
References
1.Bauman,A.,RisselC,GarrardJetal.,Cycling:GettingAustraliaMoving:Barriers,Facilitators
andInterventionstoGetMoreAustralianPhysicallyActiveThroughCycling.2008.Accessed
July,2011.Availableathttp://www.cyclingpromotion.com.au/content/view/333/145/.
2.Hamer,M.andY.Chida,Activecommutingandcardiovascularrisk:Ameta‐analyticreview.
Preventivemedicine,2008.46(1):p.9‐13.
3.Andersen,L.B.,etal.,All‐causemortalityassociatedwithphysicalactivityduringleisuretime,
work,sports,andcyclingtowork.Archivesofinternalmedicine,2000.160(11):p.1621.
4.Hu,G.,etal.,Occupational,commutingandleisure‐timephysicalactivityinrelationto
coronaryheartdiseaseamongmiddle‐agedFinnishmenandwomen.Atherosclerosis,2007.
194(2):p.490‐497.
5.BassettJr,D.R.,etal.,Walking,cycling,andobesityratesinEurope,NorthAmerica,and
Australia.JournalofPhysicalActivityandHealth,2008.5(6):p.795‐814.
6.Litman,T.andE.Doherty,TransportationCostandBenefitAnalysis:Techniques,Estimates
andImplications.VictoriaTransportPolicyInstitute,Victoria,Canada.BritishColumbia,
Canada,2009.
7.Cozens,P.M.,G.Saville,andD.Hillier,Crimepreventionthroughenvironmentaldesign
(CPTED):areviewandmodernbibliography.PropertyManagement,2005.23(5):p.328‐356.
8.Pucher,J.andR.Buehler,WalkingandCyclingforHealthyCities.BuiltEnvironment,2010.
36(4):p.391‐414.
9.Pucher,J.andL.Dijkstra,Promotingsafewalkingandcyclingtoimprovepublichealth:
lessonsfromtheNetherlandsandGermany.Americanjournalofpublichealth,2003.93(9):
p.1509.
10.Chong,S.,etal.,Relativeinjuryseverityamongvulnerablenon‐motorisedroadusers:
Comparativeanalysisofinjuryarisingfrombicycle–motorvehicleandbicycle–pedestrian
collisions.AccidentAnalysis&Prevention,2010.42(1):p.290‐296.
11.Watson,L.andM.Cameron,Bicycleandmotorvehiclecrashcharacteristics.2006,Monash
UniversityAccidentResearchCentre:Melbourne.AccessedOct,2009.Availableat
http://www.monash.edu.au/muarc/reports/muarc251.pdf
12.PopulationHealthDivision,ThehealthofthePeopleofNewSouthWales‐Reportofthe
ChiefHealthOfficer.NSWDepartmentofHealth:NorthSydney.AccessedJuly,2011.
Availableat:www.health.nsw.gov.au/publichealth/chorep/.
13.Sikic,M.,etal.,BicyclinginjuriesandmortalityinVictoria,2001–2006.MedJAust,2009.
190(7):p.353‐6.
14.Bauman,A.E.andC.Rissel,Cyclingandhealth:anopportunityforpositivechange?The
MedicalJournalofAustralia,2009.190(7):p.347‐348.
15.Lott,D.andD.Lott,Effectofbikelanesontenclassesofbicycle‐automobileaccidentsin
Davis,California..JournalofSafetyResearch1976.8:p.171‐179.
16.Watchel,A.andD.Lewis,Riskfactorsforbicycle‐motorvehiclecollisionsatintersections.
1994,InstituteofTransportationEngineers.p.30‐35.
17.Franklin,J.,TwodecadesoftheredwaycyclepathsinMiltonKeynes.Trafficengineeringand
control,1999.40(7).
18.OECD,SafetyOfVulnerableRoadUsers.1998,ScientificExpertGroupontheSafetyof
VulnerableRoadUsers(RS7),DirectorateForScience,TechnologyandIndustry.
19.AustralianBicycleCouncil,TheAustralianNationalCyclingStrategy2011‐2016.2010,
Austroads:Sydney.
11
20.NSWGovernment,NewSouthWalesBikePlan.2010,Premier'sCouncilofActiveLiving:
Sydney.
21.CityofSydney,CycleStrategyandActionPlan2007‐2017.2007,CityofSydney:Sydney.
22.Aluja,A.,M.Kuhlman,andM.Zuckerman,DevelopmentoftheZuckerman‐Kuhlman‐Aluja
PersonalityQuestionnaire(ZKA‐PQ):AFactor/FacetVersionoftheZuckerman‐Kuhlman
PersonalityQuestionnaire(ZKPQ).JournalofPersonalityAssessment,2010.92(5):p.416‐
431.
23.Rohrmann,B.RiskAttitudesScales:ConceptsandQuestionnaires,2002.AccessedOct2009.
Availablefrom:http://www.rohrmannresearch.net/pdfs/rohrmann‐ras‐report.pdf.
24.Reynolds,C.C.O.,etal.,Theimpactoftransportationinfrastructureonbicyclinginjuriesand
crashes:areviewoftheliterature.EnvironmentalHealth,2009.8(1):p.47.
25.Elvik,R.andA.B.Mysen,Incompleteaccidentreporting:meta‐analysisofstudiesmadein13
countries.TransportationResearchRecord:JournaloftheTransportationResearchBoard,
1999.1665(‐1):p.133‐140.