• No results found

Comparison of driving simulator performance with real driving after alcohol intake: A randomised, single blind, placebo-controlled, cross-over trial

N/A
N/A
Protected

Academic year: 2022

Share "Comparison of driving simulator performance with real driving after alcohol intake: A randomised, single blind, placebo-controlled, cross-over trial"

Copied!
8
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

ContentslistsavailableatSciVerseScienceDirect

Accident Analysis and Prevention

jo u r n al hom e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / a a p

Comparison of driving simulator performance with real driving after alcohol intake: A randomised, single blind, placebo-controlled, cross-over trial

Arne Helland

a,b,∗

, Gunnar D. Jenssen

c

, Lone-Eirin Lervåg

c

, Andreas Austgulen Westin

a

, Terje Moen

c

, Kristian Sakshaug

c

, Stian Lydersen

d

, Jørg Mørland

e

, Lars Slørdal

a,b

aDepartmentofClinicalPharmacology,St.OlavUniversityHospital,Trondheim,Norway

bDepartmentofLaboratoryMedicine,Children’sandWomen’sHealth,NorwegianUniversityofScienceandTechnology,Trondheim,Norway

cSINTEFTechnologyandSociety,DepartmentofTransportResearch,Trondheim,Norway

dTheRegionalCentreforChildandAdolescentMentalHealth,NorwegianUniversityofScienceandTechnology,Trondheim,Norway

eNationalInstituteofPublicHealth,Oslo,NorwayandInstituteofClinicalMedicine,UniversityofOslo,Norway

a r t i c l e i n f o

Articlehistory:

Received19September2012 Receivedinrevisedform 20December2012 Accepted28December2012

Keywords:

Automobiledriving Trafficaccidents Ethanol

a b s t r a c t

Thepurposeofthisstudywastoestablishandvalidateadrivingsimulatormethodforassessingdrug effectsondriving.Toachievethis,weusedethanolasapositivecontrol,andexaminedwhetherethanol affectsdrivingperformanceinthesimulator,andwhethertheseeffectsareconsistentwithperformance duringrealdrivingonatesttrack,alsoundertheinfluenceofethanol.Twentyhealthymalevolunteers underwentatotalofsixdrivingtrialsof1hduration;threeinaninstrumentedvehicleonaclosed- circuittesttrackthatcloselyresembledruralNorwegianroadconditions,andthreeinthesimulator withadrivingscenariomodelledafterthetesttrack.Testsubjectswereeithersoberortitratedtoblood alcoholconcentration(BAC)levelsof0.5g/Land0.9g/L.Thestudywasconductedinarandomised,cross- over,single-blindfashion,usingplacebodrinksandplacebopillsasconfounders.Theprimaryoutcome measurewasstandarddeviationoflateralposition(SDLP;“weaving”).Eighteentestsubjectscompleted allsixdrivingtrials,andcompletedatawereacquiredfrom18subjectsinthesimulatorand10subjects onthetesttrack,respectively.Therewasapositivedose–responserelationshipbetweenhigherethanol concentrationsandincreasesinSDLPinboththesimulatorandonthetesttrack(p<0.001forboth).In thesimulator,thisdose–responsewasevidentalreadyafter15minofdriving.SDLPvalueswerehigher andshowedalargerinter-individualvariabilityinthesimulatorthanonthetesttrack.Mostsubjects displayedasimilarrelationshipbetweenBACandSDLPinthesimulatorandonthetesttrack;however,a fewsubjectsshowedstrikingdissimilarities,withveryhighSDLPvaluesinthesimulator.Thismayreflect thelackofperceiveddangerinthesimulator,causingrecklessdrivinginafewtestsubjects.Overall,the resultssuggestthatSDLPinthedrivingsimulatorisasensitivemeasureofethanolimpaireddriving.The comparisonwithrealdrivingimpliesrelativeexternalvalidityofthesimulator.

© 2013 Elsevier Ltd.

1. Introduction

Impaireddriving caused byethanol and/or drugsis a major causeoftrafficaccidents,andthusamajorpublichealthproblem (Blomberg etal.,2009).Therelationshipbetweenblood ethanol concentrations (BAC) and accident risk is well established in largeepidemiologicalstudies(Borkensteinetal.,1974;Blomberg etal., 2009).With theexception ofcannabis (Ramaekers etal., 2004),similarrelationshipshavenotbeendemonstratedforother

Correspondingauthorat:DepartmentofClinicalPharmacology,St.OlavUniver- sityHospital,Pb3250Sluppen,NO-7006Trondheim,Norway.Tel.:+4772829100;

fax:+4772829110.

E-mailaddresses:arne.helland@legemidler.no,arnehelland@yahoo.no (A.Helland).

psychoactivedrugs anddrugsof abuse.Case–controlstudieson non-alcoholdrugsrequirescreeningandquantificationofalarge numberofpotentiallyimpairingdrugs,aswellasalargenumber ofcases,aseachdrughasarelativelylowprevalenceofdetection incar crashdrivers.Such studieshaveseldombeenperformed, leavingtherelationbetweenblooddrugconcentrationsandcrash risklargelyunknown.Also,bloodsamplingfordrugtestingofcon- trols–ascomparedtosimplebreathtestsinethanolstudies–is necessary,andmakestherecruitmentofcontrolsmore difficult (Versteretal.,2009a).Furthermore,post-mortemdrugconcentra- tionchangesoccurtoalargerdegreeinnon-alcoholdrugs,making interpretationoftoxicologicaldatafromstudiesofkilleddrivers difficult.

Epidemiologicalapproaches cannotestablish causalrelation- ships,andarefraughtwithmethodologicaldifficulties,including

0001-4575© 2013 Elsevier Ltd.

http://dx.doi.org/10.1016/j.aap.2012.12.042

CC BY-NC-ND license.

CC BY-NC-ND license.

Open access under

Open access under

(2)

behaviour)ofthethree“corelevels”ofdriving(Walshetal.,2008).

Muchoftheon-roadexperimentshavesofarbeenconductedinThe Netherlandsonflat,straightmulti-lanemotorways;adrivingsce- nariothatmaynotreflectconditionselsewhere.Also,legalissues andsafetyconsiderationsmayhinderon-roadexperiments,and thecostsofsuchexperimentsmaybeprohibitive.

Experimental studies utilising driving simulators may avoid someoftheproblemslistedabove.However,even verysophis- ticated simulatorscannot fullyreplicate realdriving conditions (Versteretal.,2004;Shechtman etal.,2009).Driving simulator studiesofeffectsofdepressantdrugsondrivingabilityfrequently yieldinconclusiveresultsdue tothelackofvalidation againsta knownpositivecontrol;inpractice,ethanol.Thepositivecontrol isnecessarytoensurethatcorrelationsbetweendrugintakeand drivingrelatedoutcomemeasuresactuallyreflectadrugrelated impairmentofdrivingability,andnotsimplyrandomlyobserved correlationswithnorelevancetoimpairment(Walshetal.,2008).

Ethanolasapositivecontrolalsoensuresthattheexperimental designissufficientlysensitivetotheimpairingeffectsofdepres- santdrugs.Anothercommonlimitationofdrivingsimulatorsisthe lackofvalidationagainstarealdrivingscenario;i.e.,theexternal validity.Thisleavesdoubtastowhethertestsubjectperformance inthesimulatedscenariomaypredictperformanceinrealdriving situations.

Wewantedtodevelopavalidandfunctionaltoolforassessing drugeffectsondrivingperformance,takingintoaccounttherec- ommendationsmadeinthe guidelinesforresearch ondrugged driving.Toachievethis,weconductedavalidationstudyoftheSIN- TEFdrivingsimulator.Thepurposeofthestudywastoestablisha drivingsimulatortestbatterythatissensitivetoethanoleffects, andtovalidatethetestbatterybycomparingperformanceinthe simulatorwithactualdrivingperformanceonaclosed-circuittest trackresemblingruraldrivingconditions.Eventhoughbothsimu- latorandclosedcircuitdrivingconstituteexperimentalconditions, whichdonotfullyreproducethereallifedrivingexperience,both arewidelyusedforassessingdrivingperformance,andrealdriv- ingisgenerallyconsideredtobethereferencemethodologyasfar asvalidityisconcerned.Inthispaperwepresentresultsfromthe primaryoutcomemeasureSDLP,measuredinthesimulatorandon thetesttrack.

2. Materialsandmethods 2.1. Testsubjects

Twentyhealthy,Caucasian,malevolunteersaged25–35years (mean28.7years)whohadbeeninpossessionofadriver’slicense foratleast5years(mean10.6years),wereincludedinthestudy.

Theywereallrecreationalusersofalcohol,andasagroupdrove slightlyless andhad a somewhathighereducationallevel than thegeneralpopulation.Womenandnon-Caucasianswereexcluded becauseoftheteratogenicriskassociatedwithethanoluseinthe

Fig.1.Outlineoftrialtestdesign.

formergroup,andthepossibilityofdeviantethanolmetabolism inthelatter.Theotherexclusioncriteriawerepreviousorpresent drugoralcoholabuseoratypicalreactionstoalcohol,previoushis- toryofdrivingundertheinfluence,significantadversereactionsto previousbloodsampling,regular(daily)intakeofanyprescribed drug,orhighlikelihoodofmotionsicknessasassessedwithamodi- fiedversionoftheApfelriskscoreforpostoperativevomiting(Apfel etal.,1998).Eachparticipantunderwentascreeningforeligibility, receivedwrittenandoralinformationaboutthestudyandprovided awrittenconsenttoparticipate.Thestudywasapprovedbythe RegionalEthicsCommittee,andwasregisteredasaclinicaltrialin theClinicalTrials.govdatabase.Allparticipantsreceivedagiftcer- tificateworthNOK1000(approx.USD150)uponcompletionofthe study.

2.2. Trialdesign

The experiment was designed as a randomised, placebo- controlled, single blind, crossover study. Only the necessary personnelwereinformedaboutwhichinterventionsweregiven.An outlineofthedesignispresentedinFig.1.Eachparticipantunder- wentthreedrivingtestsof1hduration,bothonaclosed-circuittest trackandinanadvanceddrivingsimulator,onsixdifferenttestdays withwashoutperiodsofminimumtwodaysbetweentestdaysto allowthedissipationofanylearningorfatigueeffects.Thedriving scenariointhesimulatorwasmodelledtomimicthetesttrack,as illustratedinFig.2,toensurethatthedrivingexperiencewould beassimilaraspossibleinthetwotestconditions.Beforetest- ingcommenced,thestudysubjectsundertookatrainingsession, bothonthetesttrackandinthesimulator,inordertofamiliarise themselveswiththetestingscenarioandminimisetheimpactof possiblelearningeffects.Ontestdays,theparticipantswereobliged todeliveraurinesampleonarrivalatthetestsitetoexcludethe presenceofdrugs.Thesubjects’weightwasregisteredeachtest day,afterwhichtheywereadministeredaweight-adjusteddoseof ethanol(0,0.7and1.05gperkgbodyweight),calculatedtoobtain anintendedbloodalcoholconcentration(BAC)duringtestingof 0,0.5and0.9g/Lonthethreedifferenttestdaysbothinthesim- ulatorandonthetesttrack,respectively.TheWidmarkequation (AndréassonandJones,1995),wasusedtoestimatetheethanol doses,assumingatotalbodywatertototalbodymassratioof0.68, abioavailabilityof75%,andametabolicrateforethanolof0.15g/L perhour.Weusedvodkamixedwithfruitextracts,orangeandlime juicetomakethedrinkspalatable.Theplacebodrinkswerespiked withnon-alcoholicvodkaflavourinwatertomimicthevodkataste.

Thedrinkswereservedinclosedplastic containers,fromwhich theparticipantswereinstructedtosipthedrinkthroughastraw.

Toavoidanobviousethanoltaste,nodrinkswerestrongerthan 10%(v/v)ethanol,andtheywerekeptcoldbytheadditionofice.

(3)

Fig.2.Exampleofthedriver’svisualimpressionontheclosed-circuittesttrack(left)andinthedrivingsimulator(right).

Theparticipantswereallowed1htofinishtheirdrinks,afterwhich theywaitedanother30minbeforethedrivingteststarted,toallow forabsorptionoftheadministeredethanol.Theorderinwhichthe participantsweretestedatdifferentBAClevelswasrandomisedby useofacounterbalanced,multi-conditiondesign.Thesameorder ofBAClevelswasusedforeachparticipantbothonthetesttrack andinthesimulator.Asanadditionalconfoundertoenhanceblind- ing,thestudysubjectswereadministereda placebopill,which theyweretoldmayormaynotcontainasedativedrug,withthe drink.Venousbloodsamplesweredrawnimmediatelybeforeand aftereachdrivingsession,andthemeanvaluewasusedasthebest estimateofthemeanBACduringtesting.

2.3. Realdrivingontesttrack

Thetesttrackdrivingwasundertakenduringafrost-freeperiod ofsixweeksintheautumn.Allstudysessionsweredoneafternight- fall,between20:00and01:00h.Thetesttrackcircuitwas1.37km long,closedtoordinarytraffic,andlaidoutinhillyterrain,withboth gentleandsharpercurves.Thetrackwashard-surfaced,withtwo laneseachapprox.2.75mwide,andhadmidlineandsidemarkings similartostandardNorwegianroadmarkings.Thus,thetesttrack closelyresembledroadstypicalofruralNorway.Surpriseobstacles (1m3foamrubbercubes)wereplacedintwolocationsontwoocca- sions,oneatthebeginningandonetowardstheendofeachdriving trip,andweretobeavoidedbythetestsubjects.Stoplightspresent intwolocationsturnedredononeoccasionduringeachtrip.The participantsdroveaninstrumentedcar(VolvoV702.4s)withauto- matictransmission,fittedwithadoublesetofpedals.Theywere instructedtodriveastheywouldnormallydoonaregularroad.A professionaldrivinginstructorwaspresentinthefrontpassenger seatduringallsessionsoftesttrackdriving,inordertointervene ifnecessary.Aphysicianwaspresentonthesiteatalltimesdur- ingtestdrives.Permissiontocarryoutthetesttrackdrivingwas grantedfromthelocalpolice.Toenablecontinuousrecordingof lateralpositionintheroadlane,thetestcarwasequippedwithan infraredwide-anglecamerafixedtotheroofofthecar,andpointing atadownwardangletotherearofthecar.Thedatawerestoredina databaseandanalysedinaprogramforphotoanalysis(OpenSource ComputerVisionLibrary).Afilteringalgorithm(Houghtransforma- tion)wasusedtoidentifyroadsidemarkings.Thecaralsofeatured otherequipmentforrecordingthelocationofthecaronthetestcir- cuit(globalpositioningsystem;GPS),speed,pedaluseandsteering wheelmovements.

2.4. Drivingsimulator

Testinginthedrivingsimulatortookplaceinlateautumnafter thetest trackdrivingtests werecompleted.Testsessionswere

doneatthesametimesduringtheeveningandnightasonthe test track, using a virtualmodel of the test track and a night- timescenario(Fig.2),toensurecomparableresultsandeliminate differences in circadianinfluences. In addition toobstaclesand stoplights,thesimulatorscenarioalsoincludedtwoincidents(a carabruptlyenteringtheroadandapedestriancrossingtheroadin frontofthedriver)thateachoccurredonceattheendofthedriving session.Thesimulatorhadtheappearanceofaregularcar(Renault Scenic)withautomatictransmissionandoriginalcontrols(Fig.3).

Informationfromtheuseofsteeringwheel,pedals,transmission etc.wasfedintoadedicateddrivingscenariographicscomputer.

Thedrivingscenariowasdepictedonscreenscovering180ofthe driver’sforwardfieldofvisionand90 oftherearfieldofvision, andsynchronouslyininternalandexternalmirrors.Thevertical fieldofviewwas47bothtothefrontandtotherear.Thesimu- latorreproducedrealisticmotion,vibrationandsoundthrougha three-axismovingplatform,avibrationsysteminthechassisand afour-channelsoundsystem.Dataonlateralposition,speed,pedal useandsteeringwheelmovementsovertheentiredurationofthe testsessionswereextracteddirectlyfromthesimulatorcomputer andlogged20timespersecond.AdetaileddescriptionoftheSINTEF simulatorcanbefoundinEngen(2008).

2.5. Measurements

The predefinedprimary outcome measurewas thestandard deviation of lateralposition (SDLP),which is a measure of the degreeofweavingofthecarontheroad.SDLPhasbeenshown tocorrelatewithBAClevelsinadosedependentmanner,andis athoroughlyvalidatedmeasureofthedegreeofdrivingimpair- ment (Verster etal., 2004).Secondary outcomemeasures were numberofbrakepedalpressuresperlap,numberofaccelerator

Fig.3. Setupofthedrivingsimulator.Vehicleandsurroundingfrontalscreens.

(4)

icationanddrivingperformance.

Bloodethanolconcentrationswerequantifiedusingaheadspace gas chromatography–mass spectrometry (GC–MS) method. In brief,200␮Lbloodwasmixedwith50␮Linternalstandard(d6- ethanol). Samples were left for 30min to achieve equilibrium beforethegasfractionwasaspiratedintoanAgilentHP6890-5973 GC–MSsystem(Agilent,PaloAlto,CA).Separationwasperformed ona J&W Scientific123-9134 DB-ALC1(30m×1.2mm)column withaheliummobilephaseandaruntimeof0.90min.Ethanolwas monitoredatm/z31andtheinternalstandardatm/z33.Thelevelof quantification(LOQ)was2mmol/L(approx.0.09g/L).Between-day coefficientofvariation(CV)calculatedfromqualitycontrolsamples was4.5%at5mmol/L(0.22g/L)and1.8%at50mmol/L(2.2g/L).

2.6. Statisticalanalyses

Anapriorisamplesizeestimationperformedwithone-tailed, pairedt-testsindicatedthatatotalsamplesizeofn=11wouldbe sufficienttodetectsignificantdifferencesinBAClevelinfluence onSDLPwithsignificancelevel (˛)of 0.05and power(1−ˇ)of 0.95.Althoughtheoretically11subjectswouldsuffice,wechoseto include20subjectsinthestudy,toallowfortheuncertaintyinthe underlyingassumptionsofthesamplesizeestimation,aswellas thepossibilityofdropouts,forinstanceduetosimulatorsickness.

Intheresultsanalyses,weusedalinearmixedmodelwithSDLP asdependentvariable, measuredBACas covariate,and partici- pantasrandomeffect.Separateanalyseswereperformedfortest trackandsimulator.Reportedresultsarefromrestrictedmaximum likelihoodestimation.Themaximumlikelihoodestimationdidnot alwaysconverge.Theindependentvariablestestedforsignificance wereBAClevel,curved/straightsectionandpartoftripdriven(each tripwasdividedinfourequalpartsof15min).Toidentifypossi- blelearningeffectsthatcouldinterferewiththeresults,theimpact ofthenumberoftripsdrivenbeforetheactualonewasalsoana- lysed.Two-sidedp-values<0.05wereconsideredsignificant.The analyseswereperformedinSPSS18andStata12.

3. Results

Ofthe20participantsenrolledinthestudy,allcompletedthree drivingsessionsonthetesttrack,while18outof20completed allthreesessionsinthedrivingsimulator.Twosubjectsdidnot completethesimulatortesting;one becauseofintolerablenau- sea,andtheotherbecauseofasurgicalprocedureunrelatedtothe study.Onthetesttrack,10outofthe60drivingsessionsdidnot yieldsufficientSDLPdatatobeincludedintheanalyses.Thecar- mountedcamerawasoutofpositionineightsessions,thecamera wasnotswitchedoninoneinstance,and oneparticipantinhis firstsessionmisinterpretedtheinstructionstodriveinlane.Thus, acompletesetofoutcomedatawasobtainedfrom10participants onthetesttrackand18participantsinthesimulator.Datafromthe validdrivingsessionsofallsubjectswereincludedintheanalyses.

3.1. Safetyandadverseevents

Nosafetyviolationsorseriousorunexpectedadverse events occurredduringthestudy.Themostcommonadverseeventinthe simulatorwasnausea,whichisaknowndisadvantageofdriving simulators.Six subjects(fouratBAC0and twoatBAC0.5)had toterminatetheirfirstsimulatorsessionearlybecauseofthis,but fiveofthemwereeventuallyabletocompleteallthreesessions.

Thus, onlyone subjecthad towithdrawfromthe studydueto nausea.Priorexperiencesuggeststhatethanolmayprotectagainst simulatorsickness,andrepeatedexposurestothesimulatortend toattenuatethenausea.Therefore,inordertopreventdropouts, allparticipantswhoterminatedtheirsessionsearlyduetonausea weretestedatthehighestBAClevelinthesubsequentsession.The randomorderwasalsomodifiedinanadditionalthreesubjectsdue tootherpracticalcauses.Thesemodificationstotherandomisation didnotaffectconcealmentoftheinterventions,anddidnotappear tointroducesystematicbias,sincetherewasnostatisticallysig- nificantcorrelationbetweenBAClevelandthenumberofprevious testsessions(Pearsoncorrelation0.241(p=0.080)insimulatorand 0.094(p=0.477)ontesttrack).

3.2. Bloodalcoholconcentrations

TheethanolconcentrationsarepresentedinTable1.Ethanol concentrationswereslightlylowerthanintendedbothinthesim- ulatorandonthetesttrack,withconcentrationscloserto0.4g/L attheintendedlevelof0.5g/L.TheBACalsotendedtobeslightly lowerinthesimulatorthanonthetesttrack.Pairedsamplet-test showedastatisticallysignificantdifferencebetweentheBAClevels insimulatorandontesttrackforthedesignatedBAClevelof0.5g/L (p=0.041);however,themeandifferencewasonly0.039g/L.For thedesignatedBAClevelof0.9g/L,therewasnostatisticallysig- nificantdifferencebetweenBAC levelsinsimulatorand ontest track(p=0.21).Inthefollowing,ethanollevelsarereferredtoas theintendedlevels(BAC0,BAC0.5andBAC0.9,respectively).

3.3. Questionnaires

Aftereachdrivingsession,theparticipantswereaskedwhether theythoughtthedrinkandthepillhadcontainedalcoholanda sedativedrug,respectively.Mostsubjectscorrectlyidentifiedthe drinkascontaining/notcontainingethanol(in32of38placebotri- als,35of38BAC0.5trialsand37of38BAC0.9trials,respectively).

However,afewmisidentifiedtheirdrinks,andquiteafewwrongly identifiedthepillascontainingasedativedrug(in15of38placebo trials,3of38BAC0.5trialsand7of38BAC0.9trials,respectively).

ThereweresignificantcorrelationsbetweenhigherBAClevels and subjective (self reported) ratings of poorer driving perfor- mancebothinthesimulator(R=0.35,p=0.013)andonthetest track (R=0.63, p<0.001). Likewise,there wasa strongcorrela- tionbetweenhigherBAClevelsandobjective(drivinginstructor

(5)

Fig.4. Regressionanalysisoftherelationshipbetweenbloodalcoholconcentration(BAC)andstandarddeviationoflateralposition(SDLP)insimulator(left;filledcircles) andontesttrack(right;opencircles).ThecirclesrepresentindividualBACandthecorrespondingSDLPvalue.Theregressionlinesandtheir95%confidenceintervalsare shownascontinuouslinesandbrokenlines,respectively.

reported)ratingsofpoorerdrivingperformanceonthetesttrack (R=0.52,p<0.001).

3.4. SDLP

Fig.4showstheindividualSDLPvaluesatthecorresponding BAC,withtheestimated regressionlineand its95% confidence interval.Bothinthesimulatorandonthetesttrack,therewere significantpositivecorrelationsbetweenBACandSDLP(positive regressionslopewithp<0.001).Theestimatedregressionlinesfor thesimulator(Eq.(1))andthetesttrack(Eq.(2))areasfollows, withstandarderrorsfortheestimatesinparentheses:

(simulator):SDLP(cm)=29.43(±2.57)+13.20(±3.61)×BAC (1) (testtrack):SDLP(cm)=22.30(±1.89)+7.61(±1.91)×BAC (2) SDLPvalueswerehigherinthesimulatorthanonthetesttrack atbaseline(placebo)conditions(29.4cmvs.22.3cm,respectively), andshowedasteeperincreasewithincreasingBAC,asseenfrom Eqs.(1)and(2),aswellasFig.4.AsevidentfromFig.4,SDLPvariance wasalsolargerinsimulatordrivingthanintesttrackdriving.

TherelationshipbetweenBAClevelsandSDLPresultsshowa dose–responseeffect,asquantifiedbytheslopes13.20and7.61in Eq.(1)and(2).Furthermore,avisualcomparisonofSDLPresults inthesimulatorandonthetesttrackineachofthe20individual subjectsshowssimilar,positiveslopesinmostsubjects(Fig.5).

To identify possible differential effects of test duration and curved/straightsectionsonSDLP,theSDLPresultswereanalysed withrespecttotimeintervals(fourequalintervalsof15mineach), andperformanceoncurvedandstraightsectionsofthedrivingsce- nario.Inthesimulator,meanSDLPvaluesweresignificantlyhigher incurved sectionsthanin straightsections(p=0.047), whereas therewerenosuchdifferencesonthetesttrack(p=0.17).Inthe simulator,statisticallysignificantdifferencesinSDLPbetweenBAC levelswereseeninallfourtimeintervals.Onthetesttrack,the differencesinSDLPweresimilarbutlesspronounced,andmostly didnotreachsignificanceduringthefirsthalfhourofthetest.In thesimulator,therewasatrendtowardshigherSDLPvalueswith longertestduration,especiallyatthehighestBAClevel.Nosuch tendencywasevidentonthetesttrack.

Toidentifypossible learningeffects that would beexpected toreduceSDLPwiththenumberofpriortestsessions,thenum- beroftripsdrivenbeforetheactualonewasalsoanalysedasan

independentvariable.However,thishadnostatisticallysignificant correlationwithSDLPresultseitherinthesimulator(p=0.70)or onthetesttrack(p=0.66).

4. Discussion 4.1. SDLP

Ourresultsshowapositivedose–responsecorrelationbetween BAC and SDLPin the simulator and onthetest track, both for individualandmeandata.Ahighdegreeofintra-individualsim- ilarity in the BAC-correlated increase in SDLPin the simulator and on the test track, suggests that SDLP is a valid and sen- sitive measure of ethanol-induced driving impairment in the simulator.

AbsolutevaluesofSDLPwerehigherinthesimulatorthanon thetesttrack,withmeanSDLPatBAC0(soberstate)of29cmand 22cm,respectively.SDLPvaluesduringplaceboconditionsinthe simulatorwerealsoconsiderablyhigherthanthoseseeninDutch on-roaddrivingtests,wheremeanbaselineSDLPisapprox.19cm (range9–30cm)(VersterandRoth,2011).Therelativelydemand- ingdrivingscenariothatwasusedinourexperimentmayaccount fortheslightlyhigherSDLPvaluesonthetesttrackthanthoseseen duringpreviouson-roadtests.HigherabsoluteSDLPvaluesinthe simulator compared toreal drivingmay beexplained by unfa- miliarity withthedrivingexperienceinthesimulator,alackof perceiveddanger,andlackofgravitationalcuesandfeedbackthat willnormallyadjuststeering.Thisnotionisalsosupportedbythe observationthatSDLPvalueswerehigherincurvedsectionsthan instraightsectionsinthesimulator,whereassuchadifferencewas notobservedonthetesttrack.Togetherwiththemoredemand- ingdrivingscenarioinourexperiment,thismayaccountforthe considerablyhigherSDLPvaluesthanthoseseenforinstancein theDutchSTISIMsimulatoremployingamonotonoushighwaysce- nario(Metsetal.,2011b).

MosttestsubjectsshowedsimilarSDLPincreasesinthesim- ulatorandonthetesttrack.However,fromtheindividualSDLP datashowninFig.5,afewsubjectsbehavedifferently,evidenced byexcessiveSDLPvaluesinthesimulator.Forinstance,testsubject no.15hadameanSDLPexceeding1matthehighestBAClevel.This wouldcorrespondtothecarbeinglocatedmostlyoutoflaneduring thetrip,whichisinaccordancewiththeactualobservationsmade

(6)

Fig.5. IndividualSDLPdataatactualBAClevelsinsimulator(filledcircles)andontesttrack(opencircles).Fortestsubject15,theBACandSDLPvaluesatthehighestBAC levelinthesimulatorwere0.79g/Land102cm,respectively.

duringthisindividual’ssimulatordriving.Itisourexperiencefrom thepresentandearliersimulatorexperimentsthatsomepartici- pantsregardthesimulatorasakindofgameandbehavemorelike virtualrallydriversinsteadoffollowingtheinstructionstodrive appropriatelyaccordingtoconditions.Thiscanexplainthelarge discrepanciesinSDLPbetweentesttrackandsimulatorseenina fewofthesubjects.Subjectno.14attainedanunexpectedlyhigh BACathishighestBAClevelinthesimulator(1.25g/L),whichmay explainthehighSDLPobservedinthatdrivingsession.Also,we cannotexcludethepossibilitythatsomeparticipants’SDLPscores wereinfluencedbysimulatorsickness.

4.2. BAC

Meansubjectiveandobjectiveratingsofintoxicationanddriv- ingperformancecorrelatedwithBAClevelintheexpectedmanner.

ThesomewhatlowerBAClevelsinsimulatorthanontesttrackmay beduetoapossibleconditionednausearesponseinthesimulator thatcouldhavecausedretentionofstomachcontentwithdelayed ethanolabsorption.Oneparticipant(subjectno.6)wasunableto

finishhisdrinkattheintendedBAC0.9levelinthesimulator,and consequentlyacquiredalowBAC.

Mostparticipantscorrectlyidentified theirdrinkascontain- ing/notcontainingethanolandthepillascontaining/notcontaining asedativedrug,althoughquiteafewoftheparticipantsmisiden- tifiedtheplacebopill,especiallyintheBAC0trials.Thisprobably reflectsanexpectation biasin somesubjects,and indicatesthat the use of placebo pills to enhance blinding of the interven- tion in experimental trials with ethanol may be worthwhile.

Previousexperiencesuggeststhatconcealmentofethanolisdif- ficultinblindedstudiesduetothedistinctivetasteandsmelland thecharacteristicandfamiliareffectsofethanol.

4.3. Comparisonwithotherdrivingsimulatorstudiesand on-roadtests

To date, there are few other studies validating the use of drivingsimulatorsfordrugand/orethanolimpairmentresearch.

Asimulator validation studypublishedin 2009useddatafrom two separate previous studies (on-road and in simulator). The

(7)

descriptionofthesimulatortheyusedsuggeststhatitwassimi- lartotheSINTEFsimulator,butthedrivingscenarioandoutcome measureweredifferent(urbantrafficandnumberofdrivingerrors atintersectionsasassessedbyadrivinginstructor,respectively).No ethanolorotherdrugswereused.Resultsindicatedrelativevalid- ityforthesimulator,andsuggestedabsolutevalidityforthetype oferrorspertaininglanemaintenance,adjustmenttostimuliand visualscanning(Shechtmanetal.,2009).

Thereare fewprevioussimulatorstudiesusingSDLPasout- comemeasure.OnlyonestudyhasvalidatedSDLPasanindicatorof unsafedrivinginthesimulatorthatwasused.Metsetal.published avalidationstudyin2011showingtheabilityoftheSTISIMdriving simulatortodifferentiatebetweendifferentBAClevelsbasedon SDLPresults.Inthisstudy,27healthyvolunteersunderwentasim- ulatoradaptationofthestandardisedDutchon-roadtestscenario (multi-lanehighwaydrivingfor 1h).BAClevelsof0g/L, 0.5g/L, 0.8g/Land1.1g/LyieldedmeanSDLPvaluesof28.0cm,29.7cm, 33.8cmand36.3cm,respectively.Thisstudydidnotvalidatethe simulatorresultsagainstarealdrivingtest(Metsetal.,2011b).

Apartfromthis,onlytwosimulatorstudiesconcerningdrivingper- formanceafterdrugintakehavebeenpublishedusingSDLPasan outcomemeasure.Metsetal.haveinvestigatedtheeffectsofcaf- feine(givenintheformoftheenergydrinkRedBull®andcoffee, respectively)ondrivingperformanceinhealthyvolunteersintwo studiesintheDutchSTISIMsimulator,andfoundsmallbutsignifi- cantreductionsinSDLPaftercaffeineadministrationinbothstudies (Metsetal.,2011a,2012).

In2009,avalidationstudywithethanolinadivided-attention steeringsimulator(DASS)waspublished. Asthenamesuggests, thesimulatorisdesignedtomeasureabilityofdividedattention.

Accordingly,itemploysaratherartificialtestscenario,wheresub- jects must keepthe car inlane and simultaneouslyrespondto peripheralvisualstimuli.Also,thesimulatoruseddidnotresemble anormalcar.Dose-dependentimpairmentwasfoundwithhigher ethanollevels(Versteretal.,2009b).

Thestandardisedon-roaddrivingtestwithSDLPastheoutcome measuredevelopedinTheNetherlandsremainsthemethodofref- erencetoexaminedrivingimpairmentfromdrugs.Insuchtesting, BAClevelsof0.5g/Land 0.8g/LonaverageincreasesSDLPfrom placeboconditionswith2.4cmand 4.3cm,respectively(Verster andRoth,2011).Ourresultsfromthetesttrackshowslightlylarger increasesinSDLP,whereastheBAC-relatedincreasesinthesim- ulatorwereconsiderablylarger.Again,thediscrepancy between ourresultsand theDutchon-roadresultsmaybeexplainedby themoredemandingdrivingscenarioemployedinourvalidation study.

4.4. Implicationsforthevalidityandfurtheruseofthesimulator

Externalvalidityofadrivingsimulatorreferstothetestsce- nario’sabilitytoinvokesimilarreactionsinthedriversasareal drivingscenario.Validityisspecificfortheparticulartypeofsce- narioandsimulator,test, andpopulationusedinthevalidation experiments,andwillnotnecessarilybetransferabletootherdriv- ingscenarios,simulators,tests,orpopulations.Externalvalidityis absoluteifthesameeffectisinvokedtothesameextentbothin thesimulatorandintherealdrivingenvironment.Relativeexter- nalvalidityimpliesthatthereexistsatrendofchangeinthesame directionbothinthesimulatorand intherealdrivingenviron- ment,butthemagnitudeofchangeisdifferent(Shechtmanetal., 2009).

There was a large degree of similarity in the relationship betweenSDLPand BAC levelsin thesimulator and onthetest track.However,theabsolutevaluesofSDLPinthesimulatorwere consistentlyhigherthanonthetesttrack.Thus,therelative(but not theabsolute) external validity of theSINTEF simulator has

beenestablished whenvalidated againsttest track drivingin a driving scenario that is representative of the demanding rural drivingconditionsinNorway,usingethanolasapositivecontrol.

Webelievethatthis validationmaybeextendedtorealdriving undersimilarconditions;however,thisassumptionhasnotbeen proven.

In the simulator, we found consistent and significant BAC- related increases in SDLP in all time intervals when the hour-long test was divided into four 15-min time inter- vals. This suggests that the duration of the simulator test in order to reach significant results may be shortened in future studies.

4.5. Limitationsofthestudy

In ourstudy,alltest subjectswerehealthyyoungmale vol- unteers, who are not representative for the general driving population. Ourresultsmaythereforegivea somewhatinaccu- rate estimation of the impact of BAC on SDLP in the general population.

Therearethreelevelsofbehaviourrelevanttotrafficsafety:

automatic, control and executive planning behaviour (Michon, 1985;Walshetal.,2008).SDLPastheprimaryoutcomemeasure inthisstudyismainlyrepresentativefortheeffectofethanolon automatedactionsatabehaviouralcontrollevel.Outcomemeas- uresofdrivingbehaviouratmanoeuvringandstrategiclevelswill bereportedinaseparatepublication.Drivingsimulatorsmaybe especially suitable totest higher behaviourallevels likehazard avoidance, dualattention, risk taking and impulsivity, both for ethical(riskofinjury)andpractical(easeandreliabilityofmea- surements)reasons.

Weemployedasingleblinddesign,keepingtheintervention concealedfromthetestsubjectsbutnotfromthestudypersonnel orthoseresponsibleforanalysingtheoutcomedata.

Unlike some of the most advanced simulators in use, the SINTEFsimulatorallowsonlylimitedtilting(threedegreesoffree- dom).Motion-basedsimulatorswithfulltiltingtechnologymight increasetherealismofthedrivingexperience,andthusheighten theexternalvalidityofthesimulator.

Severalofthetestsubjectsexperiencednauseainthesimula- tor,whichcaused onesubjecttowithdrawfromthestudy,and mayhaveaffecteddrivingbehaviourinothers.Thisis ageneral drawback of driving simulators,which may to someextent be unavoidable,evenwhenusingscreeningproceduresincludingtest drivesbeforeenrolment.Wealsoemployedaratherchallenging drivingscenario,withmanycurvesandlongduration,whichmay haveexacerbatedtheproblemsrelatedtonausea.

Thevalidationagainstrealdrivingwasdoneonaclosedtest track.Thelength(approx.1.4km)andlayout(curvy,hard-toproad approx.5.5mwidewithmidlineandsidemarkings)ofthetesttrack ensuredthatthedrivingexperienceresembledrealdrivingonrural Norwegianroads.However,itmaybeimpossibletofullyeliminate thefeelingofanartificialsituationwhendrivingonaclosedtest track.Forsafetyreasons,adrivinginstructorwaspresentinthe passengerseatatalltimesonthetesttrack,aswellasapoliceofficer onthetesttracksite.Thismayhaveconstitutedarestrainingeffect aswellasheightenedtheattentionoftestsubjects,causingthem todrivemorecarefullyandattentivelythantheywouldotherwise havedone.

Finally,ourstudyhadalimitedsamplesize,whichgenerally increasestheriskoftypeIIerrors(i.e.,failingtodetectrealdiffer- ences).Also,missingdatafrom10of60drivingsessionsonthetest trackmayhavelimitedthestatisticalsignificanceofourfindings.

Themissingdataoccurredduetorandomincidents,andwehave noreasontobelievethisintroducedsystematicbias.

(8)

depressantdrugs.

Acknowledgements

TheresearchprojectwasfundedbyagrantfromtheResearch CouncilofNorway.

Theauthorswouldliketothankthefollowing employeesat SINTEFTechnology and Society and the Department of Clinical PharmacologyatSt.OlavUniversityHospitalwhomadevaluable contributionstotheresearchproject:ThomasEngen,DagfinnMoe, KennethSørensen,TerjeGiæver,CatoMausethagen,NinaHusby, Marianne Nordtømme, Odd Hjelkrem, Guro Haltbrekken,Anita Skogholt,TrudeKristineVaagGjerde.

WewouldalsoliketothankProf.Dr.JanRamaekersatMaas- trichtUniversity,TheNetherlands,forvaluableadviceandhelpful discussionsduringthepreparationsofthestudy.

Finally,we wouldlike toexpressourgratitudetothestudy subjectswhouncomplaininglyunderwenttheratherdemanding trials.

References

Andréasson,R.,Jones,A.W.,1995.ErikM.P.Widmark(1889–1945):Swedishpioneer inforensicalcoholtoxicology.ForensicScienceInternational72(1),1–14.

B.,Verster,J.C.,2011a.PositiveeffectsofRedBull energydrinkondriving performanceduringprolongeddriving.Psychopharmacology(Berlin)214(3), 737–745.

Mets, M.A., Kuipers, E., Senerpont Domis, L.M., Leenders, M., Olivier, B., Verster, J.C., 2011b. Effects of alcohol on highway driving in the STISIM driving simulator. Human psychopharmacology. [Epub ahead of print].

Michon,J.A.,1985.Acriticalviewofdriverbehaviormodels:whatdoweknow, whatshouldwedo?In:Evans,L.,Schwing,R.C.(Eds.),HumanBehaviorand TrafficSafety.PlenumPress,NewYork,pp.485–520.

Ramaekers,J.G.,Berghaus,G.,VanLaar,M.,Drummer,O.H.,2004.Doserelatedrisk ofmotorvehiclecrashesaftercannabisuse.DrugandAlcoholDependence73 (2),109–119.

Shechtman,O.,Classen,S.,Awadzi,K.,Mann,W.,2009.Comparisonofdrivingerrors betweenon-the-roadandsimulateddrivingassessment:avalidationstudy.

TrafficInjuryPrevention10(4),379–385.

Verster,J.C.,Pandi-Perumal,S.R.,Ramaekers,J.G.,DeGier,J.J.,2009a.Drugs,Driving andTrafficSafety.BirkhäuserVerlagAG,Basel,Switzerland.

Verster, J.C., Roth, T., 2011. Standard operation procedures for conducting the on-the-road driving test, and measurement of the standard devia- tionoflateralposition(SDLP).InternationalJournalofGeneralMedicine4, 359–371.

Verster, J.C., Veldhuijzen, D.S., Volkerts, E.R., 2004. Residual effects of sleep medication on driving ability. Sleep Medicine Reviews 8 (4), 309–

325.

Verster,J.C.,Wester,A.E.,Goorden,M.,VanWieringen,J.P.,Olivier,B.,Volkerts, E.R.,2009b.Novicedrivers’performanceafterdifferentalcoholdosagesand placebointhedivided-attentionsteeringsimulator(DASS).Psychopharmaco- logy(Berlin)204(1),127–133.

Walsh,J.M.,Verstraete,A.G.,Huestis,M.A.,Mørland,J.,2008.Guidelinesforresearch ondruggeddriving.Addiction103(8),1258–1268.

Referanser

RELATERTE DOKUMENTER

The connection-oriented packet data service also provides return channel functionality, which can be used by the LLC layer to send feedback reports within the reserved time slots

Keywords: gender, diversity, recruitment, selection process, retention, turnover, military culture,

3 The definition of total defence reads: “The modernised total defence concept encompasses mutual support and cooperation between the Norwegian Armed Forces and civil society in

Having the relative low data-rate of Iridium and the results from evaluating the transport protocol (TCP) used in mind, the service discovery have a relative poor performance..

The dense gas atmospheric dispersion model SLAB predicts a higher initial chlorine concentration using the instantaneous or short duration pool option, compared to evaporation from

This report documents the experiences and lessons from the deployment of operational analysts to Afghanistan with the Norwegian Armed Forces, with regard to the concept, the main

Based on the results from Soeters’ (1997) study of cross-cultural differences in a military sample, the current study asked whether members of the military really are different

The reaction time studies conducted in the driving simulator were compared to real life measurements, previous research, and measurements of reaction time in a video- based