Incident analysis methods - ehsdb

Introduction: The models used in accident investigation can typically be grouped into three types: sequential, epidemiological, and systemic models. Incidentanalysismethods Home EHSGuidelines Downloads CaseStudies Webinars ContactUs Search Incidentanalysismethods: Introduction: Themodelsusedinaccidentinvestigationcantypicallybegroupedintothreetypes:sequential,epidemiological,andsystemicmodels.Althoughthesequentialandepidemiologicalmodelshavecontributedtotheunderstandingofaccidents;theyarenotsuitableforclarifyingthecomplexitiesanddynamicsofmodernsociotechnicalsystems. Inthesesystems,theinteractionsandeventsareconnectedincomplicatedways,andstandardsafetyengineeringtechniquesalonearenotsufficienttocomprehendtheaccidentcausation.Whenanalyzingmajoraccidentsinprocessindustries,amoresystematicandprofessionalmodelisneededthanwhensupervisorsandworkersareinvestigatinganormalminoraccidentinasimplesetting. Whatareincidents? Anincidentisanunplannedeventorchainofeventsthatresultsinlossessuchasfatalitiesorinjuries,damagetoassets,equipment,theenvironment,businessperformanceorcompanyreputation.Anearmissisaneventthatcouldhavepotentiallyresultedintheabovementionedlosses,butthechainofeventsstoppedintimetopreventthis.Theseincidentscanbeclassifiedinallkindsofseveritiesandtypes,andthusintocategories.Investigationandcauseanalysisshouldtakethesedifferentcategoriesintoconsideration. Purposeofinvestigation: ThepurposeofthisIncidentInvestigationGuideistoprovideemployersasystemsapproachtohelpthemidentifyandcontroltheunderlyingorrootcausesofallincidentsinordertopreventtheirrecurrence. TheBureauofLaborStatisticsreportsthatmorethanadozenworkersdiedeverydayinAmericanworkplacesin2013,andnearly4millionAmericanssufferedaseriousworkplaceinjury.Andtensofthousandsaresickenedordiefromdiseasesresultingfromtheirchronicexposurestotoxicsubstancesorstressfulworkplaceconditions.Theseeventscausemuchsufferingandgreatfinanciallosstoworkersandtheirfamilies,andalsoresultinsignificantcoststoemployersandtosocietyasawhole.Manymore“nearmisses”or“closecalls”alsohappen;theseareincidentsthatcouldhavecausedseriousinjuryorillnessbutdidnot,oftenbysheerluck.Practicallyalloftheseharmfulincidentsandclosecallsarepreventable. Allincidents–regardlessofsizeorimpact–needtobeinvestigated.Theprocesshelpsemployerslookbeyondwhathappenedtodiscoverwhyithappened.Thisallowsemployerstoidentifyandcorrectshortcomingsintheirsafetyandhealthmanagementprograms. Incidentinvestigationshelpemployers: Preventinjuriesandillnesses Savelives Savemoney Demonstratecommitmenttohealthandsafety Promotepositiveworkplacemorale Improvemanagement TheNeedforaMethodology: Onefascinatingyettroublingphenomenonthatoftengoesunnoticedduringincidentinvestigationsisthatindividualbeliefsplayamajordeterminingroleintheoutcome,becausethoseoutcomesdependpartlyonwheretheinvestigatorsbelievetherootcauselies.Thisincludesopinionsandassumptionstheinvestigatorhasatthebeginningalongwithprejudicesformedalongtheway.Itgoesevendeeperthanindividualbeliefsabouthowaccidentshappen,however.Evenexperiencedandwell-trainedsafetyprofessionalssometimesarebiasedinthattheysupportcausationmodelswhichmaynottellthefullstoryofhowincidentscameabout. Frequently,aninvestigationteam’sbiascanresultinnarrowly-focusedcauseanalysis.Thisshortcomingcanresultinavenuesofinvestigationbeingleftunexplored,whichcanhappenwheninvestigationteamsareinfluencedbytheirpreconceivednotions,orjumptoconclusions. Rarelyisasinglefactortoblameinisolationforaneventthatresultsinseriousharm.Animportantlessonisthat“disastersareveryrarelytheproductofasinglemonumentalblunder.”Thoroughrootcauseanalysisoftenuncoverssurprisingresults,whichunderscoresthereasonlettingpreconceivednotionsguidetheprocessmayresultinneverfindingtherealcause.Worseyet,failingtomitigatetheactualrootcauseputslivesandassetsatcontinued,preventablerisk.Indeed,there’sprevailingandpersistentmisinformationintheindustry,causingmanyEHSprofessionalstomismanagetheirincidentinvestigationinitiative. Gettingpastassumptions: Manypractitionersaremisguidedbyapersistentmythintheindustry.Themyththat’spersistedforatleasthalfacenturyisthatworkerscommittingunsafepracticesarethecauseofmostincidentsintheworkplace.Ithasbeenstated2thatthecausesofindustrialaccidentscouldbebrokendowninthisway: 88%causedby“unsafeactsofpersons” 10%causedby“unsafemechanicalorphysicalconditions” 2%unpreventable Believeitornot,manyoftheincidentinvestigationsperformedeventodayarepermeatedbythis88-10-2formula.It’sstill taughtineducationalprogramsatuniversitiesandisthereforeextremelyentrenchedintheincidentmanagementcommunity.Duetolackofknowledgeandbadbeliefs,manyEHSProfessionalsarenotusingeffectivecausationmodels. ThepersistenceofthistypeofmisconceptionisoneofthemajorreasonswhycompaniesfindarigorousincidentinvestigationmethodologytobeanincrediblyhelpfulandpowerfuladditiontotheirEHSprogram.Awell-definedinvestigationmethodologyassistsinvestigatorsinfindingtheaccanbeimplemented. METHODOLOGIES&MODELS: ninvestigationmethodologyishowyouthinkabout,understandandresolverootcausesofanincident.Whilesoftwarecansupporttheprocess,therightmethodologymustfirstbeselectedandimplemented. Theincidentsweinvestigate—accidentsandnearmisses—almostneverresultfromonecause.Mostoftheminvolvemultiple,interrelatedcausalfactors.Thiscomplexityshouldalsobereflectedintheinvestigationmethodologyused.Selectingtherightoneforyoursituationcanbechallenging. Beforewediveintosomeexamplemethodologies,let’slookatsomeanalyticalmodels.Thiswillhelpusthinkaboutthemethodologiesinthefollowingsections. Systemicmodels focusonthesystemsandprocessesoftheorganizationalcultureandleadershiptounderstandaccidentcausesasmismatchesorfailuresbetweenthosecomponents. Logicaltreemodels attempttoanalyzethecausesofaccidentsasasetofeventsandconditions,payingparticularattentiontothelogicalrelationshipsbetweenthem. Sequence-of-Events (Domino,orCausal-sequence)modelsevaluateaccidentsasacontinuoussetoffailuresthatsetoffachainreaction. Epidemiologicalmodels fromthemedicaltermforthespreadofdisease,investigateaccidentsasemanatingfromhiddenfailuresacrossallorganizationalcomponents,includingmanagement,procedureanddesign. Energymodel rootedinepidemiology,focusesonthetransferofenergycausinginjurytoaperson,andthereforeseekstofindwaystopreventsuchatransfer. Processmodels focusonwaysinwhichaproductionsystemcandeteriorateovertime,makingacleardistinctionbetweenasequenceofeventsandanyunderlyingcausalorcontributingfactors. Humaninformation-processingmodels analyzethesituationfromtheperspectiveofahumanoperatorandhisinteractionwithhisenvironment. EHSmanagementmodels explorethepossiblecontributingandcausalfactorsrelatedtothefailingsoftheorganizationanditsmanagement. We’llintroduceatleastonemethodologyasanexampleofeachoftheanalyticalmodelswejustlisted.Thesemodelshelpusunderstandhowaparticularmethodologycouldbeusedtoidentifythedirectcausesandcontributingfactorsofincidents.Theycanbeusedtoevaluateandultimatelyreducethenumberofdirectcausestowhichfurtheranalysiswillbeapplied. Thebestmethodologieshelpinvestigatorsbyutilizingmultiplemodelsofanalysistoensurethoroughresearchintorootcausesandcontributorycauses.Someofthemethodologiesdiscussedcouldbedescribedasdiagrammingtechniques.Thesecanprovideausefulframeworkfordevelopingevidencebysummarizingtheeventsinadiagram,whichprovidesaframeworkfordocumentingevidence,identifyingcausalfactors,andidentifyinggapsinknowledge. Diagramshelppreventinaccurateconclusionsbyexposinggapsinthelogicalsequenceofevents.Wheregapsareidentified,furtheranalysiscanuncovernecessarydetail. Thebestmethodologyforyoumightbeacombinationofseveraltoolsormethodologies.Someofthosecoveredinthefollowingsectionsarereallyacombinationoftoolsandtechniquesthatareassembledtogethertoformanewmethodology.Nothingpreventsyoufromdoingthesamethingtomakeanidealmethodologyforyourorganization SYSTEMATICCAUSEANALYSISTECHNIQUE(SCAT) TheInternationalLossControlInstitute(ILCI)developedSCATabout20yearsagoforthepurposeofoccupationalhealthandsafetyincidentinvestigations. SCATisasystemicModelfocusedonthesystemsandprocessesoftheorganizationalcultureandleadership,andisbasedonrootcauseanalysismethods.Thismethodologyprovidesachartwithaseriesofcrossreferencedcategories.Theinvestigatormustidentifytherelevantfactorsbyworkingsystematicallythroughthechartandidentifyingthecontributingfactorswithineachcategory. Issueswhichleadtoanincidentaredescribedaspointsatwhichtheorganizationlosescontroloverdeficiencies,whichinturnledtotheundesiredoutcome. Inotherwords,SCATasksinvestigatorstogobackbeforethecauseoftheproblemtowheretherootsofthatcausewereformed.Onecausemightbeinadequateleadership,forexample.ThefollowingfiguredepictsthepathwayofinvestigationintheSCATmethod.Asillustratedbelow,fivemainpoints-lackofcontrol,basiccauses,immediatecauses,incident,andloss-areusedinthistypeof systemictechnique.             SCAT:SystematicCauseAnalysisTechnique MANAGEMENTOVERSIGHT&RISKTREE(MORT) TheManagementOversightandRiskTree(MORT)isananalyticalprocedurefordeterminingcausesandcontributingfactors.Itarosefromaprojectundertakeninthe1970stoprovidetheU.S.Nuclearindustrywithariskmanagementprogramcompetenttoachievehighstandardsofhealthandsafety. MORT,alogicaltreemodel,isbasedonFaultTreeAnalysis(FTA),atopdown,deductivefailureanalysisprocedureusedtoanalyzecausesandrelatedfactorsofanundesiredstateusingBooleanlogictocombineaseriesoflower-leveleventsandprecursors. Faulttreeanalysismapstherelationshipbetweenfaults,subsystems,components,andcontrolsbycreatingalogicdiagramoftheoverallsystem. Everysufficientlycomplexsystemissubjecttofailureasaresultofoneormoreindividualcomponentsfailing. MORTusesacomprehensiveanalyticalprocedurethatprovidesadisciplinedmethodfordeterminingthecausesandcontributingfactorsofmajoraccidents.Themethodcanalsobeusedtoproactivelyevaluatethequalityofanexistingsystem. Accidentsaredefinedasunplannedeventsthatproducelosseswhenaharmfulagentcomesintocontactwithapersonorasset.Thiscontactcanoccurbecauseofafailureofpreventionorasanunfortunate,butacceptable,outcomeofariskthathasbeenproperlyassessedandassumed.Mostoftheeffortisdirectedatidentifyingproblemsinthecontrolofaworkprocessanddeficienciesinthebarriersinvolved,asin: 1.avulnerabletargetexposedto...2.anagentofharminthe...3.absenceofadequatebarriers. TheMORTmethodologyisless-usedtodayinwhole,butthechartingtechniqueisfairlycommon. SEQUENTIALTIMEDEVENTSPLOTTING(STEP): AtechniquethatcanbeusedtodepictabasictimelineofanincidentistheSequentialTimedEventPlot5alsoknownasaSTEPdiagram.Events,activities,andstatechangescanbeorganizedintoasinglediagraminasequence-ofeventsanalyticalmodel. Thetimelinecanfocusprimarilyon‘what’happened(theevents)andlessonwhythingshappened(thecauses).Thisisbecausetheremaybemultiple(interacting)causesforanyeventonthetimelineandcausesmaynotbeclosetogetherintimeorplace. Whileothermethodologiesmaybemorehelpfultoidentifytherootcausesofaccidentconsequences,STEPcanbeextremelybeneficialforunderstandingtheinteractionbetweenmultiplefactorsandoutcomes.Thetimeline-basedapproachclearlyandconciselygivesapictureofthe‘what’and‘when’toallowinvestigationteamstoworkbackwardstothe‘why’andthe‘how’. STEPisamultilinearsystemsapproachthatviewaccidentsasmultipleavenuesofcausalfactorsthatareinterrelatedandinteractwithotherfactorsthroughoutthesystemtoultimatelyleadtoanaccident. TheSTEPprocedurereliesonaworksheetthatprovidesstructure,visibility,andorganizationtodatagatheringandanalysis.Itgraphicallyrepresentsthebeginningandendofanaccidentsequence,detailingactorsandactionsovertime.Theprocedureaccommodateseventsthatoccurredatthesametime.Theseeventsallowinvestigatorstovisuallyrecreatethementalmapofasequenceofeventsanddeterminegaps. BARRIERANALYSISMETHOD: barrieranalysisprovidesastructuredmethodofevaluatingtheeventsrelatedtoasystemfailureorincident.Itislargelyanenergymodel,usedtofindandevaluatewaysofavoidingunwantedenergytransfer.Thetermbarrierencompassesawiderangeofconceptssuchaspersonalprotectiveequipment,machineguards,andalarms.Additionally,barrierscouldincludepreventativemeasuressuchasstandardoperatingprocedures,workinstructions,training,supervision,emergencyplans,safetyreviews,riskassessments,andLockout/Tagoutprocedures. BarrierAnalysismightbeginwithanalysisofthephysical,humanaction,natural,andadministrativecontrolsthatshouldbeinplace.Thesequenceofeventsareevaluatedtodeterminewhichbarriersfailedorsucceeded.Additionally,theanalystshoulddetermineifcertainbarrierscouldhavepreventedtheincidentiftheyhadbeenputintoplace. BarrieranalysisisoftenusedinconjunctionwithMORT.Morerecently,ithasbeenusedwithbowtiediagramsandevenadaptedforusewithSCAT. CHANGEANALYSISMETHOD ChangeAnalysisusesaprocessmodeltofocusonwaysinwhichaproductionsystemcandeteriorateovertime.Inthismethodofanalyzingsystemfailures,theinvestigatorsearchesforachangeorchangesinthesystemthatledtofailure.Themethodassumeschangeistheculpritbecausethesystemperformedadequatelyforaperiodoftimeandthensuddenlyfailed. Twoimportantconceptsinthismethodarethedirectionalandexponentialcharacteristicsofchange.Thedirectionalcharacteristicofchangeproposesthatifachangeismadethesystemwillcontinueinthatdirectionunlessanotherchangeismadetowardsanewdirectionorbacktowardtheoriginalstate.Theexponentialcharacteristicofchangesaysthatwhenmultiplechangesaremadetheircombinedeffectissaidtoaffectthesystemexponentiallyratherthanadditively.Thischangeanalysistechniquebeginswithacomparisonbetween thestateofthesystemjustbeforetheincidentandthestateofthesystemduringtheincident.Theanalystestablishesthedifferencesandevaluatesthecontributionsofeachdifference.Worksheetsareoftenusedtoguidetheinvestigatorusingchangeanalysistoconsiderfactorsofthechanges: What Where When Who NatureoftheTask WorkingConditions PresenceofaTriggerEvent RelevantManagerialControls Theanalystexaminestheincidentsituation,consideringcomparablenon-failurestates.Thecomparisonofthefailuresituationwithnonfailurestatesproducesalistofdifferencesbetweenthetwosituations.Atrainedinvestigatorcanusethecontrastbetweenfailureandnon-failurestatestoidentifykeydifferencesandanalyzetherolesthesedifferencesplayedintheincident’scausalfactors. INCIDENTCAUSEANALYSISMETHOD(ICAM) HumanPerformanceInvestigationProcess(HPIP)wasdevelopedfortheUSNuclearRegulatoryCommission(NRC)forusebyNRCinvestigators.Thiscomposite,ESHmanagementfocusedmethodologyiscomprisedofasuiteofsixtools: EventsandCausalFactorsCharting–Atoolusedtoplantheinvestigation,similarinsomeregardstotheSTEPdiagramsalreadycovered.Thechartcontainsthesequenceofeventsandthepotentialcausalfactorsthatledtotheaccident. SORTM–AguidetoHPIPModulesusedtoassistinvestigationplanning,factcollection,andidentificationofhumanperformancedifficultiesforrootcauseanalysis. BarrierAnalysis–Atechniquetoidentifyhumanperformancedifficultiesforrootcauseanalysis,basicallycoveredinaprecedingsection. HPIPModules–Identifieshumanperformancedifficultiesandprogrammaticsystemweaknesses.Eachmodule,suchasTrainingorProcedures,isbrokendownintoasetofNearRootCauses,suchasNoTraining.Finally,eachofthoseisbrokendownintoasetofRootCauses,suchasNoLearningObjective.Gettingtothebottomofthetreeofrelationships,theinvestigatorisguidedbyaseriesofpreformulatedquestions. ChangeAnalysis–Allowsunderstandingoftheeventandensurescompleteinvestigationandaccuracyofperceptions.Thisalsoascoveredinaprecedingsection. CriticalHumanActionsProfile(CHAP)–Similartochangeanalysis,CHAPprovidesanunderstandingoftheeventandensurescompleteinvestigationandaccuracyofperceptions.Itisanoperationallyorientedtechniquebasedonhumanfactorstaskanalysis. INCIDENTCAUSEANALYSISMETHOD(ICAM) IncidentCauseAnalysisMethod(ICAM)isbasedontheworkofJamesReason,whowasaprofessorofpsychologyattheUniversityofManchesterintheUnitedKingdom.Reason’sprimaryareaoffocuswasonunderstanding“HumanError”asitrelatedtoaccidents,andtheICAMmodelisbasedonhisSwissCheese,DefensesinDepthmodelofincidentcausation. AccordingtoReason,veryfewunsafeactsresultinactualdamageorinjury.Variouslayersofdefensecanonlybebreachedbytheadverseconjunctionofseveraldifferentcausalfactors. AfoundationalconceptofICAMisinevitabilityofhumanerror.Humanfactorsresearchhasshownthathumanerrorisanormalcharacteristicofhumanbehavior,andalthoughitcanbereduced,itcannotbecompletelyeliminated.ICAMisdesignedtoensurethattheinvestigationisnotrestrictedtotheerrorsandviolationsofoperationalpersonnel. Ithelpsidentifyotherfactorsthatcontributedtoanincidentinordertoidentifyrootcauses,andmakerecommendationsoncorrectiveactionstopreventrecurrence. MANYOTHERAVAILABLEMETHODS Thereareseveralpopularmethodologiesavailablewhichrequirelicensefees,andarestrictlycontrolledundercopyright.Therefore,useofsuchamethodologymightlimityouroptionswhenobtainingtrainingforyourstaff,andyourabilitytomodifyorevolvethemethodologyforyourparticularindustryororganization.However,ifyouarelookingforanoutoftheboxsolution,thesemethodologiesarefairlycommonandwell-respectedbymanyorganizations. TapRooT REASON KelvinTOP-SET                                                          Stillothertoolsareprevalentandnotsostrictlycontrolled.Considertheseifyouwantsomereadilyavailabletraining,yetalsotheflexibilitytomodifythemethodologyforyourownuniquecircumstances. Someexamplesinclude The“5Why”Method Man,TechnologyandOrganization(MTO)Analysis FunctionalResonanceAccidentModel(FRAM) HumanFactorsAnalysisandClassificationSystem(HFACS) SystemsTheoreticAccidentModelingandProcesses(STAMP) AcciMap SafetyOccurrenceAnalysisMethodology(SOAM) AccidentEvolutionandBarrierFunctionMethod(AEB) SafetyFunctionAnalysis(SFA) TripodDelta Decadesofresearchhaveprovidedanumerousmodelsandmethodologiestochoosefrom.Incidentinvestigatorshavesuchawidearrayoftoolsavailable,itcanbechallengingtofindtherightones.Wehopethishelpsencouragethem Althoughwe’veonlyprovidedasingleexamplemethodologyofeachmodeltype,therearemanymoretoresearchandevaluate.Whendesigningorselectingamethodology,itiswisetoresearchandevaluateseveral. Selectingasuitableaccidentinvestigationtechnique: Itisessentialthatworkplaceshaveaplanabouthowtoinvestigateaccidents.Irrespectiveofthetechnique,itisimportantthatthosepersonswhoareinvolvedinaccidentinvestigationknowhowtoconducttheinvestigationandareawareoftheguidelinesforinvestigatingaccidentsintheirworkplace.Thepersonswhoparticipateintheseinvestigationsshouldbenamed(usuallysafetymanagersandsupervisors)andinaddition,aworkerfromtheaccidentscenemaybeneficiallybeincludedintheinvestigation. Whenselectingasuitabletechniqueforaccidentinvestigation,thereshouldbeatleastonepersonwhohasagoodknowledgeaboutthedifferentaccidentinvestigationtechniquessuitableforuseintheirworkenvironment,andwhoisabletochoosethepropermethodforeachcase.Someminoraccidentsmaynotneedtobeinvestigatedinthesamekindofdepthasthosethathaveledtoseriousinjuries. Conclusions: Accidentsandalsonearmissesalmostneverresultfromonesinglecause,mostaccidentsinvolvemultiple,interrelatedcausalfactors.Allactorsordecision-makersinfluencingthenormalworkprocessmightalsoinfluenceaccidentscenarios,eitherdirectlyorindirectly.Thiscomplexityshouldalsobereflectedintheaccidentinvestigationprocess.Theaimofaccidentinvestigationsshouldbetoidentifytheeventsequencesandall(causal)factorsinfluencingtheaccidentscenarioinordertobeabletoproposeriskreducingmeasureswhichmaypreventfutureaccidents. Often,accidentinvestigationsinvolveusingasetofaccidentinvestigationmethods.Eachmethodmighthavedifferentpurposesandmaymaketheirowncontributiontothetotalinvestigationprocess.Itisimportanttorememberthateverypieceofapuzzleisassignificantastheothers. Graphicalillustrationsoftheeventsequenceareusefulduringtheinvestigationprocessbecausetheyprovideaneffectivevisualaidthatsummarizeskeyinformationandprovidesastructuredmethodforcollecting,organizingandintegratingcollectedevidencetofacilitatecommunicationbetweentheinvestigators.Graphicalillustrationsalsohelptoidentifyinformationgaps. Duringtheinvestigationprocess,differentmethodsshouldbeusedinordertoanalyzeemergingproblemareas.Thereshouldbeatleastonememberofthemulti-disciplinaryinvestigationteamwhohasgoodknowledgeaboutthedifferentaccidentinvestigationmethods,andisabletochoosetheoptimalmethodsforanalyzingthedifferentproblems. ClickthebelowlinkstodownloadToolboxtalk,PPTandguidelinesforAccidentinvestigationtechnique Toolboxtalk-AccidentInvestigationResponsibilities PowerPoint presentation-AccidentInvestigation PowerPointpresentation-AccidentinvestigationTechniques AccidentInvestigationtechniqueguidelines Tweet Subscribetoourmailinglist