Consistency of FMEA used in the validation of analytical procedures .pdf

《Consistency of FMEA used in the validation of analytical procedures .pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

JournalofPharmaceuticalandBiomedicalAnalysis54(2011)592–595ContentslistsavailableatScienceDirectJournalofPharmaceuticalandBiomedicalAnalysisjournalhomepage:www.elsevier.com/locate/jpbaShortcommunicationConsistencyofFMEAusedinthevalidationofanalyticalproceduresM.T.Oldenhofa,∗,J.F.vanLeeuwena,b,M.J.Nautaa,c,D.deKastea,Y.M.C.F.Odekerken-Romboutsa,M.J.Vredenbregta,M.Wedaa,D.M.BarendsaaNationalInstituteforPublicHealthandtheEnvironment(RIVM),Bilthoven,TheNetherlandsbMedicinesEvaluationBoard,TheHague,TheNetherlandscNationalFoodInstitute,DanishTechnicalUniversity(DTU),Søborg,DenmarkarticleinfoabstractArticlehistory:InordertoexploretheconsistencyoftheoutcomeofaFailureModeandEffectsAnalysis(FMEA)inReceived20May2010thevalidationofanalyticalprocedures,anFMEAwascarriedoutbytwodifferentteams.ThetwoteamsReceivedinrevisedformappliedtwoseparateFMEAstoaHighPerformanceLiquidChromatography–DiodeArrayDetection–Mass13September2010Spectrometry(HPLC–DAD–MS)analyticalprocedureusedinthequalitycontrolofmedicines.EachteamAccepted17September2010wasfreetodefinetheirownrankingscalesfortheprobabilityofseverity(S),occurrence(O),anddetectionAvailableonline20October2010(D)offailuremodes.WecalculatedRiskPriorityNumbers(RPNs)andweidentifiedthefailuremodesabovethe90thpercentileofRPNvaluesasfailuremodesneedingurgentcorrectiveaction;failuremodesKeywords:fallingbetweenthe75thand90thpercentileofRPNvalueswereidentifiedasfailuremodesneedingAnalyticalvalidationFMEAnecessarycorrectiveaction,respectively.Team1andTeam2identifiedfiveandsixfailuremodesneedingReproducibilityurgentcorrectiveactionrespectively,withtwobeingcommonlyidentified.OfthefailuremodesneedingRiskanalysisnecessarycorrectiveactions,aboutathirdwerecommonlyidentifiedbybothteams.TheseresultsshowinconsistencyintheoutcomeoftheFMEA.Toimproveconsistency,werecommendthatFMEAisalwayscarriedoutunderthesupervisionofanexperiencedFMEA-facilitatorandthattheFMEAteamhasatleasttwomemberswithcompetenceintheanalyticalmethodtobevalidated.However,theFMEAsofbothteamscontainedvaluableinformationthatwasnotidentifiedbytheotherteam,indicatingthatthisinconsistencyisnotalwaysadrawback.©2010ElsevierB.V.Allrightsreserved.1.IntroductionTheuseofFMEAinthedevelopmentofanalyticalprocesseswaspreviouslydescribedbyBormanetal.[3].AproposedanalyticalOurlaboratorycarriesoutanalysisonsuspectedillegalprocedurewassubjectedtoFMEA,wherebytheoutcomeenablesmedicinesinordertodetectpharmaceuticalcrime.Astheresultshighriskstobecontrolledoreliminated.Dejaegheretal.[4]appliedoftheseanalysesareusedincourt,thereliabilityofourresultsisFMEAtotheHPLCofadrugsubstancetoidentifythemostimpor-ofutmostimportance.Theusualwayoftestingananalyticalpro-tantfactorsaffectingthecapabilityoftheassay.FMEAwasalsocedureforreliabilityistoperformananalyticalvalidation.Thisappliedinaclinicallaboratorytoidentifyandestimatetherisksanalyticalvalidationcoversaccuracy,precision,reproducibility,offailureofsomeanalyticalprocesses.Capunzoetal.[5],andVanrepeatability,intermediateprecision,specificity,detectionlimit,Leeuwenetal.[6]demonstratedthepotentialofFMEAtoidentifyquantificationlimitandlinearity,takingalltechnicalandinstru-humanriskfactorsinaNear-Infrared(NIR)analyticalprocedure,mentalparametersintoaccount[1].However,accordingtoKiefferenablinganimprovementinthereliabilityoftheoutcomeofthat[2],“Frequentlythestepsintheprocesswhichinvolvehumanprocedurethroughrelativelysimpleinterventions.interventionaretheweaklinksintheprocess(...)QuiteofteninHowever,FMEAisalwaysbasedonsubjectivejudgements.So,validationworkthehumanelementisignored,whilemechanicalitisimportanttoknowtheconsistencyofanFMEAinordertoandtechnologicalaspectsarestudiedingreatdetail”.Riskanalysisdeterminethevalueofitsoutcome.Noliteraturewasfoundonthisthereforehasconsiderableaddedvalueinanalyticalvalidation,toaspect.assessfailuresduetohumanerror.FMEAisariskanalysistooloftenInordertostudytheconsistencyoftheoutcomeofanFMEAusedtoassesssuchfailures.whenusedinthevalidationofanalyticalprocedures,wesub-jectedaHighPerformanceLiquidChromatography–DiodeArrayDetection–MassSpectrometry(HPLC–DAD–MS)analyticalproce-dureusedinourlaboratorytoFMEAsbytwodifferentteams.We∗Correspondingauthorat:RIVM,POBox1,3720BA,Bilthoven,TheNetherlands.thencomparedtheputativefailuresofprocesselementswiththeTel.:+310302744233;fax:+310302744462.E-mailaddress:Margryt.Oldenhof@rivm.nl(M.T.Oldenhof).highestrisksidentifiedbybothteams.0731-7085/$–seefrontmatter©2010ElsevierB.V.Allrightsreserved.doi:10.1016/j.jpba.2010.09.024 M.T.Oldenhofetal./JournalofPharmaceuticalandBiomedicalAnalysis54(2011)592–595593Table1Table3ProcessstepsdefinedbyTeam1andTeam2.Scorescalesforfrequencyofoccurence(O)offailuremodesdefinedbyTeam1andTeam2.StepDescriptionTeam11Verification&validationofequipmentbytechnician2Preparingsample(s)andreferencesubstancesbytechnicianScoreDefinition3Performingmeasurementsofsample(s)andreferencesubstances10–0.001bytechnician20.001–0.0054Processingmeasurementresultsbytechnician30.005–0.025Interpretationofmeasurementresultsbytechnician40.02–0.16Verificationofidentityofpharmaceuticalsubstancebytechnician50.1–17Quantitativedeterminationofpharmaceuticalsubstance8ReportingmeasurementresultsbytechniciantoHPLC–DAD–MSexpertTeam29ReviewofthetechniciansreportbyHPLC–DAD–MSexpertScoreDefinition10ConclusionsofexaminationbyHPLC–DAD–MSexpert11Discussionofmeasurementresultsandconclusionsofexaminationby1Oncewithin5yearsorlessthan2in109HPLC–DAD–MSexpertandheadofthedepartment2Oncewithin3–5yearsor2in10912DraftingofresultofexaminationletterbyHPLC–DAD–MSexpertand3Oncewithin1–3yearsor6in107discussionofletterwithheadofdepartment4Oncewithinayearor6in10513Signingresultofexaminationletterbyheadofthedepartment5Oncewithin6months–1yearor1in10414ArchivingdossierbyHPLC–DAD–MSexpert6Oncewithin3monthsor3in1037Oncewithinamonthor1in102Stepsinitalic(11and12)wereexclusivelydefinedbyTeam2.8Oncewithinaweekor5in1029Oncewithin3–4daysor3in1010Morethanonceadayormorethan3in102.MethodsThetwoteamsfollowedthesameFMEAintroductioncourseancewithexperienceinlaboratoryqualitysystemsandHazardandperformedtheirFMEAindependentlyaccordingtotheinter-AnalysisandCriticalControlPoints(HACCP).ThememberswithnationalstandardforFMEA[7].Bothteamsconsistedoffourthesamebackgroundhadacomparablelevelofexperienceintheirmembers,includinganHPLC–DAD–MS-expert,aseniortechni-field.cianandaseniorpharmacist,mainlyparticipatinginthereviewEachteamseparatedtheanalyticalprocedureintosinglestepsofchemical–pharmaceuticalpartofregistrationfiles.However,aandsubsequentlyidentifiedfailuremodesassociatedwitheachdifferentpersonparticipatedineachteam.Thefourthmemberofstep,seeTable1.EachteamdefinedscoringscalesforseverityTeam1wasanexpertinMicrobiologicalRiskAssessmentsforfood,ofthatfailuremode(S),probabilityofoccurrence(O)andprob-whileinTeam2thefourthmemberwasanexpertinQualityAssur-abilityofdetectionofthatfailuremode(D),seeTables2–4.FromTable2Scorescalesforseverity(S)offailuremodesdefinedbyTeam1andTeam2.Table4Scorescalesforprobabilityofdetectionoffailuremodes(D)definedbyTeam1andTeam1Team2.ScoreDefinitionTeam11NoresultbydisfunctioningoftheapparatusorshortageofScoreDefinitionqualifiedpersonnel.Thiscausesreputationdamagefortheanalyticalcentreandpossiblelossofacustomer10.1–12Asignificanthigheramountofpharmaceuticalsubstanceis20.02–0.1reportedthanpresent.Thismayleadtoundeservedcondemnation30.005–0.02ofsuspectedpersonsasthereportedamounthasatherapeutic40.001–0.005effectwhereasthepresentamounthasnoeffect50–0.0013Asignificantloweramountofpharmaceuticalsubstanceisreportedthanpresent.ThismayleadtoapublichealthriskastheTeam2reportedamounthasnotherapeuticeffectwhereasthepresentScoreDefinitionamounthasaneffect4Falsepositive(identity).Thesampledoesnotcontainthe1Thedefectisclearlyvisibleortheproductis100%automaticallypharmaceuticalcomponentofinteresthoweveritisreportedtocontrolledwithregularlycalibrationandmaintenanceofthetheclientasdetected.Thismayleadtoundeservedcondemnationcontrolapparatusofsuspectedpersons.Asothernotdetectedpharmaceutical2Theproductis100%automaticallycontrolledsubstancesarepresentthismaycausepublichealthrisks3AqualifiedSPCprocesscontrolisusedwithCpk>133a5Falsenegative(identity).Thesamplecontainsthepharmaceutical4SPCbprocesscontrolisusedandimmediatelyactionisundertakensubstanceofinterestwithoutbeingdetectedandisthereforenotasthresholdsarecrossedreportedtothecustomer.Thiscanbedangerousforpublichealth5AformofSPCprocesscontroliscarriedoutandtheproductundergoesafinalcontroloff-lineTeam26Theproductis100%manuallycontrolledwithago/nogoorsomewayoffailurepreventionScoreDefinition7Theproductis100%manuallycontrolled1Unnoticed;norelevanteffect8Theproductiscontrolledrandomlyandreleasedbasedonzero2Failurenotnoticed;littleeffectdefects3Extraefforttoproduce,nodelay9Theproductiscontrolledrandomlyandreleasedbasedonan4Shortdelayinprocessacceptablequalitylevel5Moderatedelayinprocess10Theproductisnotinspectedorthedefectisnotdetectable6LongdelayinprocessduetocarryingoutrepairsaCpk=minimum{(upperlimittolerance−nominalvalue)/(3×standarddevia-7Rejectionofproducedproductstion),(nominalvalue−lowerlimittolerance)/(3×standarddeviation)},represent-8Customerendupwithfaultyreport/productingtheextentinwhichaprocessproduceswithintolerancelimits.Targetvaluefor9FaildoesnolongermeetlegalrulesCpk:≥1.33.10PeoplecangetseverelywoundedbSPC=statisticalprocesscontrol. 594M.T.Oldenhofetal./JournalofPharmaceuticalandBiomedicalAnalysis54(2011)592–595Table5FailuremodesleadingtourgentcorrectiveactionsidentifiedbyTeam1andTeam2.Team1StepFailuremodeSODRPNPerformingmeasurementsofsample(s)andreferenceWrongpositioningofprobe53460substancesbytechnicianQuantitativedeterminationofpharmaceuticalsubstanceCalculationsheethasbeenchangedincalculation52440programmePreparingsample(s)andreferencesubstancesbytechnicianSampleinhomogeneous54240Performingmeasurementsofsample(s)andreferenceFalsesettingofelectronicpipette52440substancesbytechnicianVerificationofidentityofpharmaceuticalsubstancebyFalselabellingofsample52440technicianTeam2StepFailuremodeSODRPNReviewofthetechnicianreportbyHPLC–DAD–MSexpertInadequatereview8410320Performingmeasurementsofsample(s)andreferencesubstancesbytechnicianWrongpipette8310240Preparingsample(s)andreferencesubstancesbytechnicianSampleinhomogeneous8310240Performingmeasurementsofsample(s)andreferencesubstancesbytechnicianPositionofsampleprobeincorrectly8310240Performingmeasurementsofsample(s)andreferencesubstancesbytechnicianFalsesettingofelectronicpipette8310240VerificationofidentityofpharmaceuticalsubstancebytechnicianFalselabellingofsample8310240Similarfailuremodesbybothteamsareshowninbolditalic.S=severity,O=occurrence,D=detection,RPN=RiskPriorityNumber.theassessmentsofeachteamofS,OandD,RiskPriorityNumberstheirFMEAonthedescriptionoftheanalyticalprocedureinits(RPNs)werecalculatedbyRPN=S×O×D.FailuremodesabovetheStandardOperatingProcedure,whereasTeam2alsovisitedthelab-90thpercentileweredefinedasfailuremodesneedingurgentcor-oratory–anapproachknownaswalk-through[3]–andtookthatrectiveactions;failuremodesbetweenthe75thand90thpercentilevisitintoconsiderationwhendefiningtheirFMEA.weredefinedasfailuremodesneedingnecessarycorrectiveactions.WeconcludethattheconsistencybetweenFMEAoutcomesAfterwards,wecomparedthefailuremodesidentifiedbythetwocanbeimprovedbydevelopingtheskillsofasmallnumberofteamsinbothcategories.experiencedfacilitatorswhocanhelpanalyststouseFMEAmoreeffectivelyandconsistentlybyassistinginthedefinitionofthefail-3.Resultsuremodesandratingoftheseverity,probabilityanddetectabilityindex.Moreover,theinvolvementofexperiencedFMEAfacilitatorsItwasnotedthatinbothteamstheinputoftheonewillprovevaluablewhenevaluatingtheeffectsofthecorrectiveHPLC–DAD–MSexpertinthatteamhadamajorinfluenceontheactionsthathavebeenundertaken.scores.Team1andTeam2identifiedatotalof56and60fail-Asecondpossibilitytoimprovetheconsistencyoftheoutcomeuremodesrespectively.Team1andTeam2identifiedfiveandsixofanFMEAofananalyticalprocedureistoensurethatthereisfailuremodesneedingurgentcorrectiveaction,respectively,withalwaysmorethanonetechnicalexpertintheteam,inviewoftwobeingcommonlyidentified.Theresultsofthecomparisonarehis/hermajorimpactonthescores.AtleasttwotechnicalexpertsshowninTable5.NotethatTeam1haddefinedrankingscalesforshouldbeincludedinanFMEAteam,tobalanceoutsignificantS,OandDfrom1to5,sothemaximumRPNforTeam1is125;individualdifferencesinthesecrucialjudgements.Team2haddefinedrankingscalesforS,OandDscalesfrom1toOntheotherhand,theinconsistencyintheFMEAoutcomesdoes10,sothemaximumRPNforTeam2is1000.Aboutonethirdofthenotonlyindicateadiscrepancyinthisriskanalysisprocedure.ThefailuremodesneedingnecessarycorrectiveactionwerecommonlyFMEAsofbothteamscontainedvaluableinformationthatwasnotidentifiedbybothteams(resultsnotshown).identifiedbytheotherteam.ThiscallsforFMEAstobesporadicallyperformedbyadifferentteam,wherebyeachteamisgiventhefreedomtouseaflexibleapproach.4.DiscussionTheoutcomeofthetwoFMEAsclearlyshowsinconsistency.AcknowledgementsBoththefailuremodesneedingurgentcorrectiveactionsandthefailuremodesneedingnecessarycorrectiveactionsidentifiedbyK.Hartog,R.HovingandF.Bakkerareacknowledgedfortheirthetwoteamsdifferconsiderably.Inparticular,twoofthefourcontributionincarryingouttheFMEA.failuremodesneedingurgentcorrectiveactionsofTeam2werenotidentifiedbyTeam1.ThesedifferencescanbepartiallyexplainedbythedifferentReferencesapproachchosenbytheteamswithintheFMEAframework.The[1]InternationalOrganisationforStandardizationISO,InternationalStandardFgeneralFMEAdocumentationthatwasprovidedtobothteamsISO/IEC17025,GeneralRequirementsfortheCompetenceofTestingandCalibra-includedscoretablesreferringtolargescaleindustrialproduc-tionLaboratories,InternationalOrganisationforStandardizationISO,Geneva,tionprocesses[8].Team2usedthesescaleswithoutmodifications.2005,pp.12–15.However,Team1decidedthatthesescoretableswerenotuseful[2]R.G.Kieffer,Validationandthehumanelement,PDAJ.Pharm.Sci.Technol.52(1998)52–54.foranFMEAofananalyticalprocedureanddevelopednewscore[3]P.Borman,M.Chatfield,P.Nethercote,D.Thompson,K.Truman,Theapplica-tableswiththeirowndefinitionsfortherankingsofS,OandD,tionofqualitybydesigntoanalyticalmethods,Pharm.Technol.31(2007)142–therebysignificantlyinfluencingthevaluesofS,OandD,andcon-152.[4]B.Dejaegher,M.Jimidar,M.DeSmet,P.Cockaerts,J.Smeyers-Verbeke,Y.VandersequentlytheRPNofeachfailuremodeaswellastherankingofHeyden,ImprovingmethodcapabilityofadrugsubstanceHPLCassay,J.Pharm.thesefailuremodesaccordingtotheirRPNs.Also,Team1basedBiomed.Anal.42(2006)155–170. M.T.Oldenhofetal./JournalofPharmaceuticalandBiomedicalAnalysis54(2011)592–595595[5]M.Capunzo,P.Cavallo,G.Boccia,L.Brunetti,S.Pizzuti,AFMEAClinicalLabo-[7]InternationalElectrotechnicalCommissionIEC,InternationalStandardIECratoryCaseStudy:howtomakeproblemsandimprovementsmeasurable,Clin.608122006-01,AnalysisTechniquesforSystemReliability:ProcedureforFail-LeadershipManage.Rev.(2004)37–41.ureModeandEffectAnalysis(FMEA),2nded.,InternationalElectrotechnical[6]J.F.vanLeeuwen,M.J.Nauta,D.deKaste,Y.M.C.F.Odekerken-Rombouts,M.T.CommissionIEC,Geneva,2006.Oldenhof,M.J.Vredenbregt,D.M.Barends,RiskanalysisbyFMEAasanelement[8]P.D.T.O’Connor,PracticalReliabilityEngineering,App4,JohnWileyandSonsofanalyticalvalidation,J.Pharm.Biomed.Anal.50(2009)1085–1087.Ltd.,NewYork,2002,pp.63–71.