Accident investigation techniques - OSHWiki

Selecting a suitable accident investigation technique Accidentinvestigationtechniques FromOSHWiki Jumpto:navigation,search  Share  1Vote ▼Accidentsandincidents  Accidentinvestigationandanalysis  Accidentinvestigationtechniques  Accidentsandincidents  Drivingforwork  Humanerror  Learningfromincidentsandaccidents  Nearmisses  Organisationalmeasuresofaccidentprevention  Zeroaccidentvision JormaLappalainen,PiaPerttula,FinnishInstituteofOccupationalHealth Contents 1Introduction 2Thepurposeofaccidentinvestigation 3Maintechniquesforaccidentinvestigationsandanalyses 3.1Simpletechniques 3.2Advancedtechniques 3.2.1TheAcciMap 3.2.2STAMP 3.2.3MTO-analysis 3.2.4FunctionalResonanceAccidentModel(FRAM) 4Selectingasuitableaccidentinvestigationtechnique 5Conclusions 6References 7Linksforfuturereading Introduction Themodelsusedinaccidentinvestigationcantypicallybegroupedintothreetypes:sequential,epidemiological,andsystemicmodels.Althoughthesequentialandepidemiologicalmodelshavecontributedtotheunderstandingofaccidents;theyarenotsuitableforclarifyingthecomplexitiesanddynamicsofmodernsociotechnicalsystems.Inthesesystems,theinteractionsandeventsareconnectedincomplicatedways,andstandardsafetyengineeringtechniquesalonearenotsufficienttocomprehendtheaccidentcausation.Whenanalyzingmajoraccidentsinprocessindustries,amoresystematicandprofessionalmodelisneededthanwhensupervisorsandworkersareinvestigatinganormalminoraccidentinasimplesetting. Thepurposeofaccidentinvestigation Thereareseveraldefinitionsforaccidents.Inthecontextofaccidentinvestigation,herewewillusethedefinitionthatanaccidentisanunplanned,unwanted,butcontrollableeventwhichdisruptstheworkprocessandinflictsinjuries. Anaccidentinvestigationmayhavedifferentpurposes[1]: Identifyanddescribethetruecourseofevents(what,where,when) Identifythedirectandrootcauses/contributingfactorsoftheaccident(why) Identifyriskreducingmeasurestopreventfuture,comparableaccidents(learning) Investigateandevaluatethebasisforpotentialcriminalprosecution(blame) Evaluatethequestionofguiltinordertoassesstheliabilityforcompensation(pay) Inanaccidentinvestigation,onetirestoobtainanswerstothefollowingquestions:whathappened,whyithappened,andhowcouldthishavebeenprevented? Anaccidentinvestigationshouldadoptasystematicapproachtoidentifythefactorsleadingtotheaccident,andinaddition,itshouldexaminewhatimprovementsareneededintheworkenvironmentandinorganisationalproceduresaswellasclarifyingtheresponsibilitiesofeachparticipant.Theuseofasystematicapproachconfersreliabilityontheinvestigationandmakingitpossibletodescribeinacomprehensivemannerthecourseofaccidentandallfactorsinfluencingtheaccident.Inbrief,oneneedstohaverulesofconductforaninvestigation:whoshouldparticipateandhowtoimplementtheinvestigationinpractise. Aftereveryincidentandaccident,weshoulddecidewhatkindofsafetymeasures,guiding,trainingandinformationwillbeneededintheworkplacetopreventthesamekindofincidentsandwhoshoulddealwiththisinformationinthefirstplace. Maintechniquesforaccidentinvestigationsandanalyses Simpletechniques Simpleaccidentinvestigationtechniquesdonotrequiretheuserstobeasafetyprofessional,i.e.learningthesetechniquesdonotrequirealongperiodoftrainingoracertifieddegree.Learningtoapplysimpletechniquesonlyrequiresorientationandcommitment.Atypicalfeatureforasimpletechniqueisthatthetimerequiredisnotexcessive,itshouldonlytakeacoupleofhourstoperformthiskindofaccidentinvestigation.AgoodexampleofthesesimpletechniquesistheFinnishmodelforaccidentinvestigation[2]TheFinnishmodelforaccidentinvestigationisnotstatutory,butitisapracticalandeasytousetoolforaccidentinvestigationatworkplacesthatcanbeusedbynon-experts. IntheFinnishmodelforaccidentinvestigation,itisrecommendedthattheaccidentinvestigationshouldbeconductedinworkinggroupsthatincludeindividualsfromdifferentlevelsoftheorganisation.Answersshouldbesoughttosuchquestionsas: Whathappened(description)? Wheredidithappen? Whatwerethecircumstancesattheaccidentscene? Whichpersons,machines,equipmentwereinvolvedintheaccident? Whatworkwasbeingperformedwhentheaccidentoccurred? Wasthereanythingunusualinthesituation? TheFinnishmodelforinvestigatingoccupationalaccidentsconsistsof10steps: Orientatingtotheaccidentcase:Afteranaccidenthasoccurred,itisessentialtocheckthesceneimmediatelyinordertogatherinformationaboutwhathappened.Eyewitnessesshouldbeinterviewedandcircumstancescanbephotographed.Allunusualanddevianteventsandoccurrencesshouldberecognisedandreported.Forexample,checkingthesceneoftheaccidentshouldincludethefollowingpoints: Thenamesandlocationsofvictim(s),eyewitnesses,andotherpersonswhowereworkinginthearea; Whatwasbeingdoneandwhichequipmentwasbeingused; Thecircumstancesattheaccidentscene; Thecircumstancesofthewiderworkingenvironmentingeneral(i.e.lighting,noise,andetc.); Theleveloftrainingofthepersonnelinvolved; Organisationoftheworkandresponsibilitiesofthepersonsinvolved. Describingtheeventsinchronologicalorder:Theeventsshouldbeoutlinedandseparated–Aneasywaytoundertakethisdescriptionistostartwiththeaccidentitself.Theinvestigationshouldbeextendedbackwardsuntilthelast"normal"workingactwasperformed,thusitisnotenoughtodescribeonlytotheeventthatledtotheaccident: Whatwerethepreviouseventsbeforetheaccidentoccurred? Whatwastheresult?(injurytypeandinjuredbodypart) Whatwasthetypeoftheaccident? Whatwastheconcretecauseoftheinjury? Gatheringinformationonhowthevictimwasinvolved,withthecauseoftheinjury: Thesceneandoccasion; Whatwashe/shedoingbeforetheaccidenthappened. Gatheringinformationonhowthecauseoftheinjurywasrelatedtotheaccidentcause: Thecauseoftheinjurymayexistasapartofnormaloperations,butalternativelyitmayaswellbecausedbecauseofbrokenormalfunctioningmachinesand/orequipment,orequipmentwronglyplacedintheworkplace. Gatheringinformationoncontributingfactors,i.e.whatwerethefactorsthatcontributedtotheaccident: Contributingfactors,suchasdescribedinstep2,shouldbeconsideredforeachevent; Eacheventmayincludemorethanonecontributingfactor; Recognitionofcontributingfactorsisbasedoncarefulinspectionattheactualaccidentscheme,insteadofguessingbehindtheofficedesk. Gatheringinformationonwhydidthecauseoftheinjuryexistandhowdiditcometobepresentattheaccidentscene,especiallywhenitisnotitspermanentlocation.Oneshouldalsoconsiderwhatweretheaccidentfactorscontributingtotheexistenceofthecauseoftheinjury. Consideringwaysonhowtopreventsimilaraccidentsoccurringagain. Choosingthebestmeasuresforpreventingsimilaraccidentsinthefutureandconsideringhowbesttoimplementthesemeasures: Whenseveraloptionalmeasuresexist,itisessentialtoconsiderwhichoneisthebestandmostrealisticforbeingimplemented. Choosethepersonresponsibleforimplementingthesemeasures; Setaschedulefortheimplementation. Distributinginformationontheresultsoftheaccidentinvestigationattheworkplace: Itisessentialtoinformalsootherdepartmentsinadditiontothoseatthesceneoftheaccident,becausesimilaraccidentsmayoccurinotherlocationsaswell. Followingupthatthemeasuresareimplementedandevaluatetheirimpact. Advancedtechniques GoodexamplesofmorecomplexandsystematicaccidentinvestigationtechniquesareAcciMap,modelSTAMPmodel,MTO-analysesandFRAMmethod.Eachoftheseadvancedtechniquesrequiresspecializedtrainingbeforebeingmasteredinpractice,andthereforetheywillonlybebrieflyoverviewedinthisarticle. TheAcciMap TheAcciMapaccidentanalysistechniqueisbasedonRasmussen’sriskmanagementframework[3],[4].Initially,differentaccidentscenariosareselectedandthecausalchainsofeventsareanalysedusingacause-consequencechart.Acause-consequencechartrepresentsageneralisationthataggregatesasetofaccidentalcoursesofevents.Causeconsequencechartshavebeenwidelyusedasthebasisforpredictiveriskanalysis[5].Seefigure1. Figure1.AcciMapmethod Source:[6] Thesetthatischosentobeincludedinacause-consequencechartisdefinedbythechoiceofthecriticalevent,whichreflectsthereleaseofawell-definedhazardsource,suchas“lossofcontainmentofhazardoussubstance”,or“lossofcontrolofaccumulatedenergy”.Thecriticaleventconnectsthecausaltree(thelogicrelationamongpotentialcauses)withthesubsequenteventtree.Inthisway,theAcciMapservestoidentifyrelevantdecision-makersandthenormalworksituationinwhichtheyinfluenceandmodulatepossibleaccidents. ThefocusofAcciMapisnotonthetraditionalsearchforidentifyingthe“guiltyperson”,butontheidentificationofthosepeopleinthesystemthatcanmakedecisionsresultinginimprovedriskmanagement,andhence,tothedesignofimprovedsystemsafety.[7] STAMP STAMP(SystemsTheoreticAccidentModelingandProcesses)focusesontheroleofconstraintsinsafetymanagement.Insteadofdefiningsafetyintermsofpreventingcomponentfailureevents,safetyisdefinedasacontinuouscontroltasktoimposetheconstraintsnecessarytolimitsystembehaviourtoensureonlysafechangesandadaptations.Accidentsareseenasresultingfrominadequatecontrolorenforcementofconstraintsonsafety-relatedbehaviourateachlevelofthesystemdevelopmentandsystemoperationscontrolstructures.Therefore,accidentscanbeunderstoodintermsofwhythecontrolsthatwereinplacedidnotpreventordetectmaladaptivechanges(e.g.identifyingthesafetyconstraintsthatwereviolatedateachlevelofthecontrolstructure,aswellaswhytheconstraintswereinadequateor,iftheywerepotentiallyadequate,whythesystemwasunabletoexertappropriatecontrolovertheirenforcement). Theprocessleadingtoanaccident(lossevent)canbedescribedasanadaptivefeedbackfunctionthatfailstomaintainsafetyasperformancechangesovertimetomeetacomplexsetofgoalsandvalues.Thisadaptivefeedbackmechanismallowsthemodeltoincorporateadaptationasafundamentalproperty.[8] MTO-analysis ThebasisfortheMTO-analysis(Man,TechnologyandOrganization)isthathuman,organisational,andtechnicalfactorsareequallyimportantinanaccidentinvestigation.Themethodisbasedon"HumanPerformanceEnhancementSystem(HPES)"[9],whichwillnotbedescribedindetailinthisarticle. TheMTO-analysisisbasedonthreemethods: Structuredanalysisbyuseofanevent-andcause-diagram; Changeanalysisbydescribinghoweventshavedeviatedfromearliereventsorcommonpractice; Barrieranalysisbyidentifyingtechnologicalandadministrativebarriers,whichhavefailedoraremissing. ThefirststepinanMTO-analysisistodeveloptheeventsequenceinalongitudinalsequenceandtoillustratetheeventsequenceinablockdiagram.Thenextstepistoidentifypossibletechnicalandhumancausesofeacheventandtodrawtheseverticallyintoeacheventinthediagram.Thethirdstepistoanalysewhichtechnical,humanororganisationalbarriersthathavefailedorweremissingduringtheaccidentandillustrateallmissingorfailedbarriersbelowtheeventsasshowninthefigure(Figure2.). Figure2.MTOdiagram Source:Adaptedbytheauthor[10] AchecklistforidentificationoffailurecausesisalsopartoftheMTO-methodology.Thechecklistcontainsthefollowingfactors:workorganisation,workpractice,managementofwork,changeprocedures,ergonomic/deficienciesinthetechnology,communication,instructions/procedures,education/competence,andworkenvironment.Foreachofthesefailurecauses,thereisadetailedchecklistforbasicorfundamentalcauses. FunctionalResonanceAccidentModel(FRAM) TheFunctionalResonanceAccidentModel(FRAM)andtheassociatedmethodprovideawaytodescribehowmultiplefunctionsandconditionscancombinetoproduceanadverseoutcome[11]. FRAMisbasedonthefollowingprinciples: Theprincipleofequivalenceofsuccessesandfailures.FRAMadherestotheresilienceengineeringviewthatfailuresrepresentthereversesideoftheadaptationsnecessarytocopewithrealworldcomplexityratherthanafailureofnormalsystemfunctions.Successdependsontheabilityoforganisations,groupsandindividualstoanticipaterisksandtoappreciatecriticalsituations,torecognisethemintime,andtotakeappropriateaction;failureisduetothetemporaryorpermanentabsenceofthatability. Theprincipleofapproximateadjustments.Sincetheconditionsofworknevercompletelymatchtheconditionsthathavebeenspecifiedorprescribed,individualsandorganisationsmustalwaysadjusttheirperformancesothatitcansucceedundertheexistingconditions,specificallytheactualresourcesandrequirements.Sinceresources(time,manpower,information,etc.)alwaysarelimited,suchadjustmentsareinvariablyapproximationsratherthanexactcharacterisations. Theprincipleofemergence.Thevariabilityofnormalperformanceisrarelylargeenoughtobethecauseofanaccidentinitselforeventoconstituteamalfunction.However,thevariabilityfrommultiplefunctionsmaycombineinunexpectedways,leadingtoconsequencesthataredisproportionallylarge,hencetheyproduceanon-lineareffect.Bothfailuresandnormalperformanceareemergentratherthanresultantphenomena,becauseneithercanbeattributedtoorexplainedonlybyreferringtothe(mal)functionsofspecificcomponentsorparts. Theprincipleoffunctionalresonance.Thevariabilityofanumberoffunctionsmayeverynowandthenresonate,i.e.,reinforceeachotherandtherebyleadtovariabilitysuchthatonefunctionwillexceedthenormallimits.Theconsequencesmayspreadthroughtightcouplingsratherthanviaidentifiableandenumerablecause-effectlinks,e.g.,asdescribedbytheSmallWorldPhenomenon.Thiscanbedescribedasaresonanceofthenormalvariabilityoffunctions,henceasfunctionalresonance.Theresonanceanalogyemphasisesthatthisisadynamicphenomenon,hencenotattributabletoasimplecombinationofcausallinks. WhenconductinganaccidentinvestigationwithFRAM,theexplanationisproducedbyproceedingthroughthefollowingsteps: Step1.Identifyessentialsystemfunctions,usingnormaloraccident-freeperformanceasabaseline.Thisstepcharacteriseseachfunctionseparatelybutdoesnottrytoarrangeorordertheminanyway.Thestartingpointmaybeanexistingtaskanalyses,procedures,expertknowledge,etc.Thecharacterisation involvesthefollowingsixaspects: Input(I):thatwhichthefunctionprocessesortransformsorthatwhichstartsthefunction, Output(O):thatwhichistheresultofthefunction,eitheranentityorastatechange, Preconditions(P):conditionsthatmustexistbeforeafunctioncanbeexecuted, Resources(R):thatwhichthefunctionneedsorconsumestoproducetheoutput, Time(T):temporalconstraintsaffectingthefunction(withregardtostartingtime,finishingtimeorduration), Control(C):howthefunctionismonitoredorcontrolled. Eachfunctionmaybedescribedbyasimpletable,whichcanthenbeusedforfurtheranalysis.Itisalsopossibletoshowthefunctionsgraphicallyusingahexagontorepresenteachfunction(FRAMmodules,Figure3). Figure3.AFRAMmoduledescribinganactivityorfunctionintermsofsixaspects Source:[12] Step2.Characterisetheobservedvariabilityofsystemfunctions,consideringbothactualandpotentialvariability.ThepurposeofFRAMistoprovideanexplanationoftheaccidentintermsofcombinationsofperformancevariabilities.Thesecondstepisthereforeforeachfunctiontodescribetheactualvariabilityduringtheaccident.Thismaypointtootherfunctionsthatmustbecharacterisedaspartoftheexplanation.Forinstance,iftheinputtoafunctioncametoolate,orwasofthewrongkind,thenthesourceofthatinput–i.e.,anotherfunction–mustbedescribedandcharacterised.Thismayinturnrequireevenmorefunctionstobedescribed,untilonehasaccountedforthetotalscenario. Step3.Identifyanddescribethefunctionalresonancefromtheobserveddependencies/couplingsamongfunctionsandtheobservedperformancevariability.Theoutputofthefirstandthesecondstepsisalistoffunctionseachcharacterisedbytwoormoreofthesixaspects.(Notethatafunctionmayrequireseveralinstancesofanaspecttobedescribed.)Thedependenciesamongfunctionscanbefoundbymatchingorlinkingtheiraspects.Forexample,theoutputofonefunctionmaybea)theinputtoanotherfunction,b)constitutearesource,c)fulfilapre-condition,ord)enforceacontrolortimeconstraint.Theresultisanoveralldescriptionofhowthefunctionswerelinkedorcoupledintheaccidentscenario,andthereforeadescriptionofhowfunctionalvariabilitypropagatedthroughthesystem.Ingeneral,thelinksspecifywherethevariabilityofonefunctionmayhaveanimpact,orhowitmaypropagate.Manysuchoccurrencesandpropagationsofvariabilitymaycreatearesonanceeffect:althoughthevariabilityofeachfunctionmaybebelowthenormaldetectionthreshold,incombinationtheymaybecomea‘signal’,hencethisconstitutesarisk.Thisstepmaybesupportedbyavisualisationofhowthefunctionsarelinked.Thiskindofvisualisationcanbevaluableintracingfunctionaldependencies,buttheanalysisshouldneverthelessbebasedonthedescriptionofthefunctionsratherthanonthegraphicalrepresentation. Step4.Identifybarriersforvariability(dampingfactors)andspecifyrequiredperformancemonitoring.Barriersaremeanstopreventanunwantedeventfromtakingplace,ortoprotectagainsttheconsequencesofanunwantedevent[9].Barrierscanbedescribedintermsofbarriersystems(theorganizationaland/orphysicalstructureofthebarrier)andbarrierfunctions(themannerbywhichthebarrierachievesitspurpose).Thefourfundamentalbarriersystemsare: Physicalbarriersystemsthatblockthemovementortransportationofmass,energy,orinformation; Functionalbarriersystemsthatsetuppre-conditionsthatmustbemetbeforeanaction(byahumanand/ormachine)canbeundertaken; Symbolicbarriersystemsthatareindicationsofconstraintsonactionthatarephysicallypresent;and Incorporealbarriersystemsthatareindicationsofconstraintsonactionthatarenotphysicallypresent. Inadditiontorecommendationsforbarriers,aFRAManalysiscanprovidethebasisforissuingrecommendationsonhowtomonitorperformanceinordertodetectexcessivevariability.Performanceindicatorsmaybedevelopedbothforfunctionsandforthecouplingsbetweentheindicators. Selectingasuitableaccidentinvestigationtechnique Itisessentialthatworkplaceshaveaplanabouthowtoinvestigateaccidents.Irrespectiveofthetechnique,itisimportantthatthosepersonswhoareinvolvedinaccidentinvestigationknowhowtoconducttheinvestigationandareawareoftheguidelinesforinvestigatingaccidentsintheirworkplace.Thepersonswhoparticipateintheseinvestigationsshouldbenamed(usuallysafetymanagersandsupervisors)andinaddition,aworkerfromtheaccidentscenemaybeneficiallybeincludedintheinvestigation. Whenselectingasuitabletechniqueforaccidentinvestigation,thereshouldbeatleastonepersonwhohasagoodknowledgeaboutthedifferentaccidentinvestigationtechniquessuitableforuseintheirworkenvironment,andwhoisabletochoosethepropermethodforeachcase.Someminoraccidentsmaynotneedtobeinvestigatedinthesamekindofdepthasthosethathaveledtoseriousinjuries. Somebasicpracticalguidanceoninvestigatinganaccidentcanbefoundinthepublication:Investigatingaccidentsandincidents.[13] Conclusions Accidentsandalsonearmissesalmostneverresultfromonesinglecause,mostaccidentsinvolvemultiple,interrelatedcausalfactors.Allactorsordecision-makersinfluencingthenormalworkprocessmightalsoinfluenceaccidentscenarios,eitherdirectlyorindirectly.Thiscomplexityshouldalsobereflectedintheaccidentinvestigationprocess.Theaimofaccidentinvestigationsshouldbetoidentifytheeventsequencesandall(causal)factorsinfluencingtheaccidentscenarioinordertobeabletoproposeriskreducingmeasureswhichmaypreventfutureaccidents.[14] Often,accidentinvestigationsinvolveusingasetofaccidentinvestigationmethods.Eachmethodmighthavedifferentpurposesandmaymaketheirowncontributiontothetotalinvestigationprocess.Itisimportanttorememberthateverypieceofapuzzleisassignificantastheothers. Graphicalillustrationsoftheeventsequenceareusefulduringtheinvestigationprocessbecausetheyprovideaneffectivevisualaidthatsummariseskeyinformationandprovidesastructuredmethodforcollecting,organisingandintegratingcollectedevidencetofacilitatecommunicationbetweentheinvestigators.Graphicalillustrationsalsohelptoidentifyinformationgaps.[15] Duringtheinvestigationprocess,differentmethodsshouldbeusedinordertoanalyseemergingproblemareas.Thereshouldbeatleastonememberofthemulti-disciplinaryinvestigationteamwhohasgoodknowledgeaboutthedifferentaccidentinvestigationmethods,andisabletochoosetheoptimalmethodsforanalysingthedifferentproblems.[16] References ↑Sklet,S.,Methodsforaccidentinvestigation,NTU2002 ↑MerjamaJ.,ReiniläJ.-M.,MäkeläinenJ.,Vaaratilanteidenjatyötapaturmientutkinta,Työturvallisuuskeskus,2009. ↑Rasmussen,J.&Svedung.I.,ProactiveRiskManagementinaDynamicSociety.SwedishRescueServicesAgency,Karlstad,Sweden,2000. ↑Rasmussen,J.,RiskManagementinaDynamicSociety:AModellingProblem.SafetyScience,27(2/3),1997,pp.183-213. ↑Leveson,N.G.,Safeware:SystemSafetyandComputers,Addison-Wesley,Reading,MA,,1995. ↑Salmon,P.M.,Cornelissen,M.,Trotter,M.J.,Systems-basedaccidentanalysismethods:AcomparisonofAccimap,HFACS,andSTAMP,SafetyScience50,2012,pp.1158–1170 ↑Qureshi,Z.H.,AReviewofAccidentModellingapproachesforComplexCriticalSociotechnicalSystems.Command,Control,CommunicationsandIntelligenceDivisionDSTODefenceScienceandTechnologyOrganisation,EdinburghSouthAustralia5111Australia,2008.Availableat:[1] ↑Leveson,N.,ANewAccidentModelforEngineeringSaferSystems.Availableat:[2] ↑Sklet,S.,Methodsforaccidentinvestigation,NTU2002 ↑Sklet,S.,Methodsforaccidentinvestigation,NTU2002 ↑Hollnagel,E.,Barriersandaccidentprevention,Aldershot,UK,Ashgate,2004. ↑Hollnagel,E.,Barriersandaccidentprevention,Aldershot,UK,Ashgate,2004. ↑HSE:Investigatingaccidentsandincidents.Availablefrom:http://www.hse.gov.uk/pubns/hsg245.pdf ↑Sklet,S.,Methodsforaccidentinvestigation,NTU2002 ↑Sklet,S.,Methodsforaccidentinvestigation,NTU2002 ↑Sklet,S.,Methodsforaccidentinvestigation,NTU2002 Linksforfuturereading CanadianCentreforOccupationalHealthSafety,Accidentinvestigation(2006-04-20).Retrievedon10October2012,from:[3] Henderson,J.,Whittington,C.,Wright,K.,Accidentinvestigation-Thedrivers,methodsandoutcomes,HSEContractResearchreport344/2001. Hovden,J.,Albrechtsen,E.,Herrera,I.A.,Isthereaneedfornewtheories,modelsandapproachestooccupationalaccidentprevention?SafetyScience,Volume48,Issue8,October2010,pp.950-56. Learningfromaccidents(nopublishingdate).Retrievedon10October2012,from:[4] Lundberg,J.,Rollenhagen,C.,Hollnagel,E.,What-You-Look-For-Is-What-You-Find–Theconsequencesofunderlyingaccidentmodelsineightaccidentinvestigationmanuals,SafetyScience,Volume47,Issue10,December2009,pp.1297-1311. OSHACADEMY,Module7DevelopingSolutions(nopublishingdate).Retrievedon10October2012,from:[5] Reason,J.,Managingtherisksoforganizationalaccidents,AshgatePublishing,Aldershot,1997. Workplacesafetyadvice(2000-2012),Howtoinvestigateanaccidentatwork?Retrievedon10October2012,from:[6] Contributors Deroiste,PiaPerttula,[email protected] Retrievedfrom"http://oshwiki.eu/index.php?title=Accident_investigation_techniques&oldid=246951" Category:Accidentsandincidents Navigationmenu Views Page Discussion Viewform Viewsource History Personaltools Login Navigation Mainpage AbouttheOSHwiki OSHwiki-TableofContents EU-OSHAwebsite discover Recentchanges Help Semanticsearch OSHwikicommunity articles Createnewarticle NewPages Search   Tools Whatlinkshere Relatedchanges Specialpages Printableversion Permanentlink Pageinformation Citethispage Browseproperties languages English Thispagewaslasteditedon21February2017,at17:21. Privacypolicy AboutOSHWiki Disclaimers