Continuing Process Verification - 1

CONTINUED PROCESS VERIFICATION
MONITORING AND MAINTAINING

THE VALIDATED STATE
LIFECYCLE APPROACH TO PROCESS
VALIDATION FDA STAGE 3
GMP MonitoringWhat, Why, Where, and How .................................................................................................. 3
Rational Revalidation ........................................................................................................................................ 8
Dos and Donts of Control ChartingPart I ....................................................................................................... 17
The Dos and Donts of Control ChartingPart II ............................................................................................... 21
Understanding the Process Nonconformity ConceptThe Scientific Reason for Process Validation ................... 27
VariationPast, Present, and Future ............................................................................................................... 36
Having It All .................................................................................................................................................... 42
Substandard Data and Documentation PracticesCase Study #5 ..................................................................... 47
Using Probability Distributions to Make Decisions ............................................................................................ 54
Case Study #3Process Validation Failure of a Liquid Product Batch
Size IncreaseIdentical Manufacturing Tanks ............................................................................................. 67
Like-For-Like Changes: Is Validation Testing Needed? Validation Case Study #7 ............................................... 73
Understanding and Reducing Analytical ErrorWhy Good Science Requires Operational Excellence ................. 78
Meeting Specifications is Not Good EnoughThe Taguchi Loss Function .......................................................... 85
Process Modeling: A Powerful Weapon in the Fight to Reduce VariationPart 1 .............................................. 90
White Spots on TabletsCompliance Case Study #8........................................................................................ 97
Compliance Case Study #5Secondary Packages With Defective Glue Joints ................................................ 102
Gamal Amer
GMP Monitoring
What, Why, Where, and How
Gamal Amer
Welcome to Maintaining Validation. KEY POINTS DISCUSSED

This column addresses topics associated with The following key points are discussed in this
maintaining the validated state of processes, article:
equipment, facilities, utilities, and analytical t.POJUPSJOHJTBOJNQPSUBOUBTQFDUPG
methods (i.e., anything that has been validated or NBJOUBJOJOHWBMJEBUJPO
qualified and must be continually maintained in t.POJUPSJOHJODMVEFTJOGPSNBUJPODPMMFDUJPO
the validated state). We intend this column to be a PWFSUJNF SFWJFXJOHBOEBOBMZ[JOHUIF
useful resource for daily work applications. JOGPSNBUJPO BOEEFGJOJOHXBSOJOHMJNJUT
This column was suggested by attendees at the t.POJUPSJOHJTBQQMJDBCMFUPQSPDFTTQBSBNFUFST
14th International Validation Week Meeting held in SBXNBUFSJBMRVBMJUZ VUJMJUZQFSGPSNBODF
Philadelphia in October of 2008. The general topic QSPEVDUDIBSBDUFSJTUJDT BOEPUIFSJNQPSUBOU
of maintaining validation was a subject of great considerations
interest. An ongoing discussion addressing these t5IF64'PPEBOE%SVH"ENJOJTUSBUJPO
questions and associated topics is warranted and draft guidance Process Validation: General
will be useful to our readers. Principles and Practices (1) identifies continued
Validation and compliance practitioners are QSPDFTTWFSJGJDBUJPO JODMVEJOHNPOJUPSJOH BT
well aware that the conformance lots of process BOJNQPSUBOUBTQFDUPGNBJOUBJOJOHWBMJEBUJPO
validation, or the IQ/OQ/PQ of equipment, are at t.POJUPSJOHJOUIFESVHBOENFEJDBMEFWJDF
most snapshots in time. These events serve to JOEVTUSJFTJTBSFHVMBUPSZSFRVJSFNFOU
document standard performance when completed. t.POJUPSJOHJTBMTPHPPEFOHJOFFSJOHQSBDUJDF
However, they do not guarantee continued t$SJUJDBMQPJOUTXJUIJOUIFPQFSBUJPOXIFSF
good performance. It is through monitoring, QSPCMFNTBSFMJLFMZUPPDDVSTIPVMECF
data evaluation, statistical analysis, change NPOJUPSFE
control, and other programs that ongoing good t.POJUPSJOHSFRVJSFTDBMJCSBUFEJOTUSVNFOUT
performance is maintained. These topics, including SFHVMBSJOUFSWBMT EBUBDPMMFDUJPO EBUBBOBMZTJT
relevant examples, are addressed in Maintaining BDDFQUBCMFMJNJUT BOEEFGJOFEBDUJPOTUPCF
Validation. taken for excursions
Reader comments, questions, and suggestions t4FMFDUJPOPGQBSBNFUFSTUPNPOJUPSNBZCF
are needed to help us fulfill our objective for this CBTFEPOQSPEVDUSFRVJSFNFOUT SFHVMBUPSZ
column. Case studies submitted by readers are most SFRVJSFNFOUT NBOVGBDUVSJOHFYQFSJFODF BOE
welcome. We need your help to make Maintaining other factors
Validation a useful resource. Please send your t.POJUPSJOHJTBLFZQBSUPGBUPUBMQSPHSBNUP
comments and suggestions to journal coordinating ensure an ongoing state of control.
editor Susan Haigney at shaigney@advanstar.com.
[
For more Author
ABOUT THE AUTHOR
information,
Gamal Amer, Ph.D., is principal at Premier Compliance Services, Inc., management consultants
go to
for compliance and manufacturing operations performance in the life sciences industry. He holds a
gxpandjvt.com/bios
Ph.D. in chemical engineering and has over 27 years of experience in the pharmaceutical and related
industries. He can be reached by e-mail at vpainc@aol.com or amerg@premierc.net.
PROCESS VALIDATION Continued Process Verification 3

Gamal Amer
INTRODUCTION: WHAT IS t21 CFR 211.42-10(iv). i#VJMEJOHTBOE

MONITORING? 'BDJMJUJFTw %FTJHOBOE$POTUSVDUJPO'FBUVSFT
"DDPSEJOHUP8JLJQFEJB iUPNPOJUPSPSNPOJUPSJOH
"TZTUFNGPSNPOJUPSJOHFOWJSPONFOUBM
HFOFSBMMZNFBOTUPCFBXBSFPGUIFTUBUFPGBTZTUFNw conditions.
8FCTUFSTVHHFTUTUIBUNPOJUPS W
iJTUPXBUDI t211.110 (a).i4BNQMJOHBOE5FTUJOHPG
PSDIFDLPOBQFSTPOPSBUIJOHw JNQMZJOHEBUB *O1SPDFTT.BUFSJBMTBOE%SVH1SPEVDUTw

DPMMFDUJPOBOESFWJFX
XIJMFNPOJUPS O
iJTPOF i4VDIDPOUSPMQSPDFEVSFTTIBMMCFFTUBCMJTIFE
XIPXBSOTPGGBVMUTPSJOGPSNTPGEVUZPSDPVME UPNPOJUPSUIFPVUQVUBOEUPWBMJEBUFUIF
CFTPNFUIJOHUIBUSFNJOETPSXBSOTw JNQMZJOH QFSGPSNBODFPGUIPTFNBOVGBDUVSJOHQSPDFTTFT
BOUJDJQBUJPOBOEDPOUSPM
#BTFEPOUIFTFEFGJOJUJPOT UIBUNBZCFSFTQPOTJCMFGPSDBVTJOHWBSJBCJMJUZ
NPOJUPSJOHFOUBJMTUIFGPMMPXJOHUISFFEJTUJODU JOUIFDIBSBDUFSJTUJDTPGJOQSPDFTTNBUFSJBMBOE
BDUJWJUJFT UIFESVHQSPEVDUw
t$PMMFDUJOHJOGPSNBUJPOPWFSUJNF t211.122 (h).i1BDLBHJOHBOE-BCFMJOH
t3FWJFXJOHBOEBOBMZ[JOHUIFJOGPSNBUJPO $POUSPMw
i.BUFSJBMTFYBNJOBUJPOBOEVTBHF
DPMMFDUFE UIVTDPOUJOVBMMZCFJOHBXBSFPGUIF DSJUFSJB1SJOUJOHEFWJDFTyTIBMMCFNPOJUPSFE
current state UPBTTVSFUIBUBMMJNQSJOUJOHDPOGPSNTUPUIF
t%FGJOJOHSFNJOEFSTBOEXBSOJOH QSJOUTQFDJGJFEJOUIFCBUDIQSPEVDUJPOSFDPSEw
t21 CFR 820.70 (a).i1SPEVDUJPOBOE1SPDFTT
*UJTJNQPSUBOUUPSFDPHOJ[FUIBUDPMMFDUJOH $POUSPMTw
(FOFSBMi&BDINBOVGBDUVSFS
EBUBBOEGJMJOHJUXJUIPVUSFWJFXBOEBOBMZTJT TIBMMEFWFMPQ DPOEVDU DPOUSPM BOENPOJUPS
JTDPOTJEFSFEHBUIFSJOHBOEOPUNPOJUPSJOH QSPEVDUJPOQSPDFTTFTUPFOTVSFUIBUBEFWJDF
.POJUPSJOHJNQMJFTHBUIFSJOHEBUBPWFSBMPOHQFSJPE DPOGPSNTUPJUTTQFDJGJDBUJPOTw
PGUJNFBUQSFFTUBCMJTIFEJOUFSWBMT PSXIFODFSUBJO t820.70 (2).i1SPEVDUJPOBOE1SPDFTT
FWFOUTPDDVS5IJTFOBCMFTUIFVOEFSTUBOEJOHPG $POUSPMTw
i.POJUPSJOHBOEDPOUSPMPG
XIBUJTHPJOHPO BOUJDJQBUJPOBOEDPSSFDUJPOPG QSPDFTTQBSBNFUFSTBOEDPNQPOFOUBOEEFWJDF
QSPCMFNT BOEFOIBODFEDPOUSPMPGUIFPQFSBUJPO DIBSBDUFSJTUJDTEVSJOHQSPEVDUJPOw
'SPNBHPPENBOVGBDUVSJOHQSBDUJDF (.1
QPJOU t820.75 (b).i1SPDFTT7BMJEBUJPOw
i&BDI
PGWJFX NPOJUPSJOHJNQMJFTDPMMFDUJPOPGJOGPSNBUJPO NBOVGBDUVSFSTIBMMFTUBCMJTIBOENBJOUBJO
BCPVUUIFQFSGPSNBODFPGBTZTUFNSFMBUFEUP(.1 QSPDFEVSFTGPSNPOJUPSJOHBOEDPOUSPMPG
NBOVGBDUVSJOH4VDITZTUFNWBSJBCMFDPVMECFB QSPDFTTQBSBNFUFSTGPSWBMJEBUFEQSPDFTTFTUP
SBXNBUFSJBM JOUFSNFEJBUF VUJMJUZTZTUFN GSFRVFODZ FOTVSFUIBUUIFTQFDJGJFESFRVJSFNFOUTDPOUJOVF
PGEFWJBUJPOT DPOEJUJPOTXJUIJOUIFNBOVGBDUVSJOH UPCFNFUw
GBDJMJUZ QSPEVDUDIBSBDUFSJTUJDT PSQFSTPOOFMSFMBUFE t820.75 (b2).i1SPDFTT7BMJEBUJPOw
i'PS
JTTVFT5IFJOGPSNBUJPODPMMFDUFEJTUIFOSFWJFXFE WBMJEBUFEQSPDFTTFT UIFNPOJUPSJOHBOEDPOUSPM
BOEBOBMZ[FEUPEFUFSNJOFXIFUIFSPSOPUUIF NFUIPETBOEEBUB UIFEBUFQFSGPSNFE BOE
TZTUFNCFJOHNPOJUPSFEJTJOBTUBUFPGDPOUSPM XIFSFBQQSPQSJBUF UIFJOEJWJEVBM T
QFSGPSNJOH
4IPVMEUIFEBUBTVHHFTUBQPUFOUJBMQSPCMFNCBTFE UIFQSPDFTTPSUIFNBKPSFRVJQNFOUVTFETIBMM
POQSFEFUFSNJOFEWBMVFT JOWFTUJHBUJPODBOUIFOCF CFEPDVNFOUFEw
initiated. tICH Q7, Good manufacturing Practice Guide
5IF'%"ESBGUHVJEBODFProcess Validation: for Active Pharmaceutical Ingredients (9). This
General Principles and Practices (1) identifies three EPDVNFOUEJTDVTTFTWBSJPVTNPOJUPSJOHUPQJDT
TUBHFTJOUIFMJGFDZDMFBQQSPBDIUPQSPDFTTWBMJEBUJPO JOTFWFSBMTFDUJPOT'PSFYBNQMF TFDUJPO
5IFUIJSEPGUIFTF DPOUJOVFEQSPDFTTWFSJGJDBUJPO TUBUFTUIBUSFHVMBSRVBMJUZSFWJFXTTIPVMECF
DBQUVSFTUIFFTTFODFPGWBMJEBUJPO.BJOUBJOJOHUIF DPOEVDUFEUPWFSJGZUIFDPOTJTUFODZPGUIF
WBMJEBUFETUBUFUISPVHIPVUPOHPJOHQFSGPSNBODF QSPDFTT*UFNTDJUFEJODMVEFDSJUJDBMJOQSPDFTT
PGUIFNBOVGBDUVSJOHQSPDFTT5IJTJTUIFTBNF DPOUSPMBOEUFTUSFTVMUT CBUDIGBJMVSFT
PCKFDUJWFGPSBMMUIJOHTWBMJEBUFEPSRVBMJGJFE DSJUJDBMEFWJBUJPOT OPODPOGPSNBODFT BOE
NBJOUBJOJOHUIFJSJOUFOEFEPQFSBUJPOBMTUBUF JOWFTUJHBUJPOTQSPDFTTDIBOHFT BOBMZUJDBM
UISPVHIPVUUIFJSMJGFUJNFPGVTF NFUIPEDIBOHFT TUBCJMJUZEBUB SFUVSOT
DPNQMBJOUT BOESFDBMMTBOEDPSSFDUJWFBDUJPOT
WHY MONITOR? 0UIFSTQFDJGJDTFDUJPOTEJTDVTTJOHNPOJUPSJOH
.POJUPSJOHJOUIFESVHBOENFEJDBMEFWJDF JODMVEF4FDUJPOT6UJMJUJFT 8BUFS
JOEVTUSJFTJTBTQFDJGJFESFHVMBUPSZSFRVJSFNFOU *O1SPDFTT4BNQMJOHBOE$POUSPMT
The following is a partial list of such regulatory and 4UBCJMJUZ.POJUPSJOHPG"1*T $MFBOJOH
HVJEBODFSFRVJSFNFOUT 7BMJEBUJPO BOE3FDPWFSZPG.BUFSJBMTBOE
4PMWFOUT
4 PROCESS VALIDATION Continued Process Verification

Gamal Amer
tICH Q8 Pharmaceutical Development

*$) t1SPDFTTEFWJBUJPOT
2GPDVTFTPOEFWFMPQNFOUPGUIFDPNNFSDJBM t"1*BUUSJCVUFTTVDIBTJNQVSJUZQSPGJMF
QSPEVDU)PXFWFS UIFOFFEGPSPOHPJOH t*OUFSNFEJBUFTRVBMJUZ
NPOJUPSJOHJTSFDPHOJ[FEBTGPMMPXTi5IF
NBOVGBDUVSJOHQSPDFTTEFWFMPQNFOUQSPHSBNNF t8BUFSTZTUFNQFSGPSNBODFVTFEUPTVQQPSUUIF
PSQSPDFTTJNQSPWFNFOUQSPHSBNNFTIPVME NBOVGBDUVSJOHPQFSBUJPO
JEFOUJGZBOZDSJUJDBMQSPDFTTQBSBNFUFSTUIBUTIPVME t&OWJSPONFOUBMDPOEJUJPOTPGUIFGBDJMJUZ
CFNPOJUPSFEPSDPOUSPMMFE FH HSBOVMBUJPOFOE surrounding the process
point) to ensure that the product is of the desired t'BDJMJUZBOEFRVJQNFOUDMFBOJOHFGGFDUJWFOFTT
RVBMJUZw"MTP i$PMMFDUJPOPGQSPDFTTNPOJUPSJOH t1FSGPSNBODFPGQSPDFEVSFTBOEQFSTPOOFM
EBUBEVSJOHUIFEFWFMPQNFOUPGUIFNBOVGBDUVSJOH t$SJUJDBMVUJMJUZPVUQVU FH XBUFSGPSJOKFDUJPO
QSPDFTTDBOQSPWJEFVTFGVMJOGPSNBUJPOUPFOIBODF XBUFSRVBMJUZ DPNQSFTTFEBJSDIBSBDUFSJTUJDT
QSPDFTTVOEFSTUBOEJOHw*$)2TVHHFTUTUIBU t.JDSPCJBMNPOJUPSJOHBOEFOEPUPYJO
DPOUJOVPVTQSPDFTTWFSJGJDBUJPO XIFSFBQQMJDBCMF
NPOJUPSJOHXJUIJOUIFGBDJMJUZ VUJMJUZ BOE
DPVMECFBOBMUFSOBUJWFBQQSPBDIUPQSPDFTT DMFBOFEFRVJQNFOU
WBMJEBUJPOJOXIJDINBOVGBDUVSJOHQSPDFTT t1FSGPSNBODFPGTVQQMJFSTWJTWJTUIFRVBMJUZ
QFSGPSNBODFJTDPOUJOVPVTMZNPOJUPSFEBOE PGUIFNBUFSJBM DPNQPOFOU TFSWJDF FUDCFJOH
FWBMVBUFE supplied
tICH Q9 Quality Risk Management
*$) t- BCPSBUPSZBOBMZUJDBMSFTVMUTBOEUIF
2SFDPHOJ[FTNPOJUPSJOHBTQBSUPGWBSJPVT QFSGPSNBODFPGBOBMZUJDBMUFDIOJRVFT
SJTLNBOBHFNFOUBDUJWJUJFT5FDIOJRVFTVTFE t"OBMZUJDBMFRVJQNFOUQFSGPSNBODF
JODMVEFGBJMVSFNPEFBOEFGGFDUTBOBMZTJT '.&"
t&RVJQNFOUNBJOUFOBODF
GBJMVSFNPEF FGGFDUT BOEDSJUJDBMJUZBOBMZTJT t.JDSPCJBMMPBETJOQSPEVDU NBOVGBDUVSJOH
'.&$"
IB[BSEBOBMZTJTBOEDSJUJDBMDPOUSPM TQBDF VUJMJUJFT FUD
QPJOUT )"$$1
BOEIB[BSEPQFSBCJMJUZBOBMZTJT
)";01
3JTLNBOBHFNFOUTIPVMECFVTFE *OTIPSU NPOJUPSJOHTIPVMEDPWFSUIFFOUJSF
UPTFMFDU FWBMVBUF BOEJOUFSQSFUUSFOESFTVMUT operation and should collect appropriate data and
BOEUPJOUFSQSFUNPOJUPSJOHEBUB2VBMJUZSJTL CFJNQMFNFOUFEBUUIFBQQSPQSJBUFGSFRVFODZ'PS
NBOBHFNFOUTIPVMEBMTPCFVTFEJOWBMJEBUJPOUP FYBNQMF JOUIFDBTFPGFOWJSPONFOUBMNPOJUPSJOH
EFUFSNJOFUIFFYUFOUGPSGPMMPXVQBDUJWJUJFTTVDIBT UIFOPSNJTUPDPMMFDUUFNQFSBUVSF SFMBUJWF
TBNQMJOH NPOJUPSJOH BOESFWBMJEBUJPO IVNJEJUZ QBSUJDVMBUFDPVOU NJDSPCJBMMFWFMT
BOESFMBUJWFQSFTTVSJ[BUJPOCFUXFFOTQBDFT5IF
5IFOFFEUPNPOJUPSJTOPUPOMZEJDUBUFECZUIF GSFRVFODJFTPGDPMMFDUJOHTVDIEBUBEFQFOETPOUIF
SFHVMBUJPOCVUBMTPCZHPPEFOHJOFFSJOHQSBDUJDFT UZQFPGBSFBCFJOHNPOJUPSFEBOEJTEFGJOFECZ
.POJUPSJOHJTVTFEUPEPUIFGPMMPXJOH DPNQFOEJBMFOUSJFT SFHVMBUPSZHVJEBODF BOEPS
t'VMGJMMUIFQSPNJTFPGDPOUJOVPVTJNQSPWFNFOU operating experience.
in that collection and analysis of operating data
JNQSPWFTUIFVOEFSTUBOEJOHPGUIFVOEFSMZJOH WHERE TO MONITOR
UFDIOPMPHZBOETVQQPSUTDPOUJOVPVTJNQSPWFNFOU .POJUPSJOHJTBMXBZTQFSGPSNFEBUUIFDSJUJDBM
t&OTVSFTZTUFNTBOEPQFSBUJPO QSPDFTT GBDJMJUZ QPJOUTXJUIJOUIFPQFSBUJPO1PJOUTXIFSF
BOEQFSTPOOFM
BSFBMXBZTJOBTUBUFPGDPOUSPM QSPCMFNTBSFLOPXOUPPDDVSPSMJLFMZUPPDDVS
DPNQMJBOUXJUI(.1 BOEJOBWBMJEBUFETUBUF and at the end of any critical steps are also logical
t%FWFMPQCBTFMJOFJOGPSNBUJPOUPFTUBCMJTI TQPUTUPNPOJUPS*ONBOVGBDUVSJOHQSPDFTTFT
PQFSBUJOHDBQBCJMJUJFT POFXPVMEBMTPNPOJUPSUIFQSPEVDUBOEUIF
t*EFOUJGZQSPCMFNTCFGPSFUIFZPDDVS conditions at the point where the product is at
t.BJOUBJOUIFWBMJEBUFETUBUFBOEEFNPOTUSBUF JUTIJHIFTUWVMOFSBCJMJUZ8IFONPOJUPSJOHB
DPOTJTUBODZ4FBTPOBMWBSJBUJPONBZCFPCTFSWFE XBUFSTZTUFN EBUBTIPVMECFDPMMFDUFEBUDSJUJDBM
steps in the process and at the critical use points.
WHAT TO MONITOR IN A GMP *OFOWJSPONFOUBMNPOJUPSJOH NJDSPCJBMBOE
ENVIRONMENT particulate data are collected at surfaces where
"MMQBSBNFUFSTLOPXOPSTVTQFDUFEUPBGGFDUUIF IVNBOPSNBDIJOFBDUJWJUJFTPDDVS BOEBUUIF
RVBMJUZ FGGJDBDZ PSTBGFUZPGUIFQSPEVDUTIPVMECF EPPST XIFSFUIFEJSFDUJPOPGBJSGMPXJTPGDSJUJDBM
NPOJUPSFE5IFTFJODMVEFUIFGPMMPXJOHFYBNQMFT JNQPSUBODFUPQSFWFOUDSPTTDPOUBNJOBUJPO
t$SJUJDBMQSPDFTTQBSBNFUFST $11
1SPDFTTFTBOEBTTPDJBUFEBDUJWJUJFTTIPVMEDBSFGVMMZ
t1SPEVDUDSJUJDBMRVBMJUZBUUSJCVUFT $2"
FWBMVBUFVTJOHSJTLBOBMZTJTQSJODJQMFTUPEFUFSNJOF
t1SPDFTTQFSGPSNBODF BQQSPQSJBUFNPOJUPSJOHSFRVJSFNFOUT

Gamal Amer
Table: Environmental sampling frequencies for DIBOHFTGPSUIFQBSBNFUFSCFJOHNPOJUPSFE PS

aseptic manufacturing facilities. CBTFEPOFYQFSJFODF
'PSFYBNQMF FOWJSPONFOUBMTBNQMJOH
Area Classification Sampling Frequency GSFRVFODJFTGPSBTFQUJDNBOVGBDUVSJOHGBDJMJUJFT
BSFEFGJOFECZUIF6OJUFE4UBUFT1IBSNBDPQFJB
Class 100 Each shift* 641

BOE'%"HVJEBODFPOBTFQUJD
QSPDFTTJOH
JOUIF5BCMF
Support to Class 100 (Class 10,000) Each shift*
Other support areas (Class
Twice/week ESTABLISHING THE PARAMETERS
100,000) TO MONITOR
Potential product/container con-
Twice/week &TUBCMJTIJOHXIJDIQBSBNFUFSTUPNPOJUPSJT
tact
OPUBUSJWJBMNBUUFS5IFQBSBNFUFSTUPNPOJUPS
Non-product contact are Once/week
(Class 100,000 or less)
NVTUCFUIFDPSSFDUQBSBNFUFSTXIJDI JGUIFZ
EFWJBUF XPVMEDBVTFQSPEVDUDIBSBDUFSJTUJDT
*FDA guidance on sterile processing suggests daily monitoring (13).
UPTVGGFS"OFYBNQMFXPVMECFNPOJUPSJOH
EFMJWFSZUJNFPGBEFWJDFDPNQPOFOUGSPN
HOW TO MONITOR UIFTVQQMJFS*UNBZCFUIBUUIFTVQQMJFS
.POJUPSJOHSFRVJSFTUIFGPMMPXJOH EFMJWFSTDPNQPOFOUTPOUJNF CVUEFMJWFSTEFGFDUJWF
t$BMJCSBUFEJOTUSVNFOUTTVJUBCMFUPDPMMFDUUIF DPNQPOFOUT4PBMUIPVHI CBTFEPOUIFNPOJUPSJOHPG
SFRVJSFEJOGPSNBUJPO EFMJWFSZUJNF JUXPVMEBQQFBSUIBUOPQSPCMFNFYJTUT
t& TUBCMJTIFE SFHVMBS JOUFSWBMTGPSDPMMFDUJOHBOE one can plainly see that the resulting product would
recording data IBWFBEFGFDUJWFDPNQPOFOU5IFNBOVGBDUVSJOH
t&JUIFSBQBQFSPSFMFDUSPOJDSFQPTJUPSZGPSUIF PSHBOJ[BUJPOXPVMEIBWFQSPCMFNTXJUIUIFQSPEVDU
EBUBUPCFDPMMFDUFE EFTQJUFUIFGBDUUIBUUIFQVSDIBTJOHPSHBOJ[BUJPOJT
t& TUBCMJTIFEHSBQIJDBMBOEPSTUBUJTUJDBM NFFUJOHJUTPCKFDUJWFT5IVTUIFQBSBNFUFSTUPNPOJUPS
UFDIOJRVFTUPGBDJMJUBUFEBUBFWBMVBUJPO NVTUTVQQPSUUIFQSPEVDURVBMJUZBOECFDPOTJTUBOU
t%BUBBOBMZTJTUPEFUFDUBOEJEFOUJGZUSFOETPS BDSPTTBMMEFQBSUNFOUT
JTTVFTUPBWPJEQSPCMFNT 4PNFPGUIFCBTJTGPSDIPPTJOHUIFQBSBNFUFSTUP
t& TUBCMJTIJOHBDUJPOMFWFMTXIFSFDPSSFDUJWF NPOJUPSJODMVEFUIFGPMMPXJOH
BDUJPOTBOEQSFWFOUJWFBDUJPOT $"1"
NBZCF t1SPEVDUSFRVJSFNFOUT TVDIBTUFNQFSBUVSF
taken TFOTJUJWJUZ QVSJUZPGUIFQSPEVDUPSJOUFSNFEJBUFT
t%FGJOJOHUIFUZQFTPGBDUJPOTUPCFUBLFOJG process yield or the yield of certain critical steps
issues or trends are identified. t3FHVMBUPSZSFRVJSFNFOUT TVDIBTUIFSFRVJSFNFOU
PGMPXNJDSPCJBMDPOUFOUGPSTUFSJMFBSFBTBOE
"NPOJUPSJOHQSPHSBNDPOTJTUTPGUIFGPMMPXJOH products
actions: t$PNQFOEJBM FOHJOFFSJOH *40BOEGFEFSBM
t4BNQMFUIFEFTJSFEQBSBNFUFSBUUIFEFTJHOBUFE TUBOEBSET TVDIBTPQFSBUJOHDPOEJUJPOTPGB
JOUFSWBMT5IFJOUFSWBMTBSFEFDJEFECZUIF CMFOEFSCBTFEPOUIFFOHJOFFSJOHEFTJHODSJUFSJB
DSJUJDBMJUZPGUIFQBSBNFUFSBOECZUIFFYQFDUFE PSJPODPOUFOUJOQVSJGJFEXBUFSCBTFEPO641
GMVDUVBUJPOXJUIJOUIFWBMVFPGUIFQBSBNFUFS QVCMJTIFEDSJUFSJB
t$ PSSFMBUFUIFWBMVFPGUIFQBSNFUFSUPUJNF t- JUFSBUVSFBOEJOEVTUSZFYQFSJFODFBCPVOET
OPSNBMMZEPOFJOUIFGPSNPGBDPOUSPMDIBSU XJUIFYBNQMFTPGXIBUPUIFSNBOVGBDUVSFST
t%FGJOFBMFSUBOEBDUJPOMFWFMTBOEXIBUZPV NPOJUPSBOEXIZ*UBMTPEFUBJMTUIFGSFRVFODJFT
XPVMEEPJGUIFZBSFSFBDIFE/PSNBMMZUIFBDUJPO and suggests how to deal with trends and issues
MFWFMTBSFEFGJOFECBTFEPOUIFEFTJHOWBMVFTPGUIF as they arise.
QBSBNFUFS XIJMFUIFBMFSUMFWFMTBSFCBTFEPOUIF
WBMVFTBUXIJDIZPVFYQFDUUPPQFSBUF
t%FGJOFJOWFTUJHBUJWFBDUJPOT FH $"1"
UPCF SUMMARY
conducted if issues arise .POJUPSJOHBOPQFSBUJPOJTB(.1SFRVJSFNFOU
t$PNNVOJDBUFSFTVMUTPGJOWFTUJHBUJPOTBOEEFWJTF *UTFSWFTUPJEFOUJGZQSPCMFNTCFGPSFUIFZIBQQFO
NFUIPEPMPHZUPDJSDVNWFOUQPUFOUJBMQSPCMFNT *UBMMPXTUIFNBOVGBDUVSFSUPUBLFOFDFTTBSZTUFQT
to ensure the process is in a state of control. The
4BNQMJOHGSFRVFODJFTGPSNPOJUPSJOHDBOCFEFGJOFE EBUBDPMMFDUFEIFMQTFTUBCMJTIBCBTFMJOFGPSUIF
CZUIFOBUVSBMDZDMFPGUIFPQFSBUJPO CZDPNQFOEJBM PQFSBUJPO4FBTPOBMWBSJBUJPOPOUIFWBMJEBUFE
SFDPNNFOEBUJPOT CZUIFBOUJDJQBUFEGSFRVFODZPG TZTUFNNBZCFEFNPOTUSBUFE$POUJOVPVT

Gamal Amer
NPOJUPSJOHBMMPXTGPSPOHPJOHWFSJGJDBUJPOBOE '%" $'3 C

i1SPDFTT7BMJEBUJPO w"QSJM
FOTVSFTUIBUTZTUFNTSFNBJOJOBWBMJEBUFETUBUF
$IPPTJOHUIFWBSJBCMFTUPNPOJUPS FTUBCMJTIJOHUIF *OUFSOBUJPOBM$POGFSFODFPO)BSNPOJTBUJPO *$)

GSFRVFODZPGNPOJUPSJOH EFWFMPQJOHUIFBOBMZUJDBM Q7, Good Manufacturing Practice Guide for Active
BOETUBUJTUJDBMBQQSPBDIUPWJTVBMJ[FUIFNPOJUPSFE Pharmaceutical Ingredients "VHVTU
QBSBNFUFS BOEEFGJOJOHUIFBMFSUBOEBDUJPOMFWFMT *$) Q8, Pharmaceutical Development .BZ
BTXFMMBTUIFBQQSPBDIUPSFTQPOEBSFUIFCBTFTPGB *$) Q9, Quality Risk Management +VOF
TVDDFTTGVMNPOJUPSJOHQSPHSBN 641 Pharmacopeia Forum 641
+BOVBS Z
'FCSVBSZ
REFERENCES '%" (VJEBODF GPS *OEVTUS Z Sterile Drug Products
'%" (VJEBODFGPS*OEVTUSZ Process Validation: General Produced by Aseptic Processing 4FQUFNCFS JVT
Principles and Practices, Draft Guidance /PWFNCFS
'%" $'3 JW
i#VJMEJOHTBOE'BDJMJUJFT w ARTICLE ACRONYM LISTING
"QSJM CAPA Corrective Action and Preventive Action
'%" $'3 B
i4BNQMJOHBOE5FTUJOHPG*O CPP Critical Process Parameters
1SPDFTT.BUFSJBMTBOE%SVH1SPEVDUT w"QSJM CQA Critical Quality Attributes
'%" $'3 I
i1BDLBHJOHBOE-BCFMJOH$POUSPM w GMP Good Manufacturing Practices
"QSJM FDA US Food and Drug Administration
'%" $'3 B
i1SPEVDUJPO BOE 1SPDFTT FMEA Failure Mode Effects Analysis
$POUSPMT w"QSJM FMECA Failure Mode, Effects, and Criticality Analysis
'%" $'3
i1SPEVDUJPO BOE 1SPDFTT HACCP Hazard Analysis and Critical Control Points
$POUSPMT w"QSJM HAZOP Hazard Operability Analysis
'%" $'3 C
i1SPDFTT7BMJEBUJPO w"QSJM USP United States Pharmacopeia

0SJHJOBMMZQVCMJTIFEJOUIF8JOUFSJTTVFPGThe Journal of Validation Technology

Hosam Aleem, Tim McCarthy, and Rodger Edwards
Rational Revalidation
ABSTRACT INTRODUCTION
5IFFDPOPNJDQSPCMFNTUIFXPSMEJTDVSSFOUMZ 5IFFDPOPNJDQSPCMFNTUIFXPSMEJTDVSSFOUMZ
GBDJOHIBWFDBVTFETFWFSFDPTUQSFTTVSFPO GBDJOHIBWFDBVTFETFWFSFDPTUQSFTTVSFPOUIF
QIBSNBDFVUJDBMNBOVGBDUVSJOH3FWBMJEBUJPOPG QIBSNBDFVUJDBMJOEVTUSZ5IFTMPXEPXOJOOFX
NBOVGBDUVSJOHQSPDFTTFTBOEFRVJQNFOUDPNQSJTFT ESVHEFWFMPQNFOUIBTFYBDFSCBUFEUIJTTJUVBUJPO
BTJHOJGJDBOUDPTUUPUIFJOEVTUSZ.BJOUBJOJOH "MUFSOBUJWFTUSBUFHJFTTVDIBTPVUTPVSDJOHSFTFBSDI
BWBMJEBUFETUBUVTJTBSFHVMBUPSZSFRVJSFNFOU BOEEFWFMPQNFOU 3%
UPVOJWFSTJUJFTBOETNBMM
5FDIOJDBMJTTVFT QBUJFOUSJTL BOECVTJOFTT CJPUFDITIBWFOPUZFUCFFOGSVJUGVM BOESFTFBSDIJO
DPOTJEFSBUJPOTJNQBDUUIFPSHBOJ[BUJPOBMBQQSPBDI NPMFDVMBSCJPMPHZJOWPMWJOHHFOPNJDT QSPUFPNJDT
UPSFWBMJEBUJPO BOENFUBCPMPNJDT DPMMFDUJWFMZUFSNFEiPNJDTw

This discussion addresses the need for IBWFOPUZFUEFMJWFSFEBOZNBKPSDPNNFSDJBM
SFWBMJEBUJPOXJUIQBSUJDVMBSBUUFOUJPOUPCVTJOFTT CSFBLUISPVHIT/PWFMPQFSBUJPOUFDIOPMPHJFT
DPOTJEFSBUJPOT4USBUFHJFTUPQSJPSJUJTFSFWBMJEBUJPO TVDIBTQSPDFTTBOBMZUJDBMUFDIOPMPHZ 1"5

BNPOHTZTUFNTBOEIPXUPEFDJEFPOXIBU IPX BOEDPOUJOVPVTQSPDFTTJOHIBWFOPUCFDPNF
BOEXIFOUPSFWBMJEBUFBSFEJTDVTTFE"MUFSOBUJWF NBJOTUSFBN3FDFOUSFHVMBUPSZGJOFTPGTPNF
BQQSPBDIFTXIFOSFWBMJEBUJPOJTOPUQPTTJCMFBSF NBOVGBDUVSFST QSJDFDPOUSPMTJO&VSPQF BOEUIF
TVHHFTUFE$PTUTBTTPDJBUFEXJUISFWBMJEBUJPO VODFSUBJOUZSFHBSEJOHIFBMUIDBSFSFGPSNTJOUIF64
BSFEFTDSJCFE"DUJWJUJFTUPSFEVDFUIFOFFEGPS DPNQPVOEUIFTJUVBUJPO
SFWBMJEBUJPOJODMVEJOHTUBUJTUJDBMQSPDFTTDPOUSPM 3FWBMJEBUJPOPGNBOVGBDUVSJOHQSPDFTTFTBOE
BSFEJTDVTTFE#VTJOFTTNPEFMTGPSBQQSPBDIJOH FRVJQNFOUDPNQSJTFTBTJHOJGJDBOUDPTUUPUIF
SFWBMJEBUJPOJODMVEJOHJOIPVTFBOEPVUTPVSDJOHBSF JOEVTUSZ.BJOUBJOJOHBWBMJEBUFETUBUVTJTB
considered. SFHVMBUPSZSFRVJSFNFOU)PXFWFS UIFFYFDVUJPO
$IBOHFNBOBHFNFOUJTBLFZDPNQPOFOUPGB PGUIJTSFRVJSFNFOUSFRVJSFTUIPSPVHIFWBMVBUJPO
SFWBMJEBUJPOQSPHSBN"MTPLFZJTBTUSVDUVSFE PGUFDIOJDBMBOECVTJOFTTDPOTJEFSBUJPOT"
BQQSPBDIUPWBMJEBUJPONBJOUFOBODFBTUJNF NBKPSJTTVFJOUIFJOEVTUSZJTBTTFTTJOHBOE
CBTFEBOEFWFOUCBTFE$POTJEFSBUJPOTJOUIFTF NBOBHJOHSJTLUPUIFQSPEVDU BTGPSFYBNQMFJO
EJTDVTTJPOTEPOPUDPNQSPNJTFQBUJFOUTBGFUZBOE UIF*OUFSOBUJPOBM$POGFSFODFPO)BSNPOJTBUJPOT
QSPEVDURVBMJUZ#VTJOFTTGBDUPSTPOMZEFUFSNJOF *$)
Q9EPDVNFOU
*OUIFDVSSFOUFDPOPNJD
UIFPSEFSJOXIJDIUIFTZTUFNTBSFUPCFSFWBMJEBUFE DMJNBUFIPXFWFS UIFTDPQFPGSJTLTIPVMECF
6OEFSOPDJSDVNTUBODFTXPVMEUIFQSPEVDUPG XJEFOFEUPJODMVEFSJTLUPUIFCVTJOFTTBTXFMM
BNBOVGBDUVSJOHQSPDFTTUIBUJTOPUWBMJEBUFE 8JUIPVUBCVTJOFTTUIFSFXPOUCFBQSPEVDUGPS
UPSFHVMBUPSZFYQFDUBUJPOTCFSFMFBTFEUPUIF XIJDIUPBTTFTTUIFSJTL5IFPSJHJOBMRVBMJUZ
NBSLFUQMBDF TUBOEBSE*40PGJTBOFYBNQMFPGB
RVBMJUZNBOBHFNFOUTZTUFNJNQMFNFOUBUJPOUIBU
[
ABOUT THE AUTHORS
For more Author Hosam Aleem has spent several years in the pharmaceutical industry as a validation and cali-
information,
bration engineer at both IPR owners and contract manufacturers. He may be reached by email at
hosam.aleem@manchester.ac.uk. Dr. Tim McCarthy is a professor of engineering design at the
go to
University of Wollongong, NSW Australia. Dr. Rodger Edwards is a senior lecturer at the School
gxpandjvt.com/bios of Mechanical, Aerospace and Civil Engineering at the University of Manchester Institute of Science
& Technology and University of Manchester.

EJEOPUUBLFFDPOPNJDGBDUPSTJOUPBDDPVOU*U QFSJPEJDDSJUJDBMSFWBMJEBUJPOUPFOTVSFUIBUUIFZ
CFDBNFBCVSFBVDSBUJDCVSEFOUIBUEJEOPUEFMJWFS SFNBJODBQBCMFPGBDIJFWJOHUIFJOUFOEFESFTVMUTw
JUTGVMMQPUFOUJBM BTJUVBUJPOUIBUXBTTVCTFRVFOUMZ 'VSUIFSNPSF "OOFYPGUIFTBNF(.1FOUJUMFE
SFDUJGJFEJOUIFWFSTJPO i2VBMJGJDBUJPOBOE7BMJEBUJPOwTUBUFT i'BDJMJUJFT
5IJTQBQFSEJTDVTTFTWBSJPVTBTQFDUTPG TZTUFNT FRVJQNFOU BOEQSPDFTTFTTIPVMECF
SFWBMJEBUJPOTQFDJGJDBMMZSFMBUFEUPCVTJOFTTDPTUT QFSJPEJDBMMZFWBMVBUFEUPWFSJGZUIBUUIFZBSF
#ZSFWBMJEBUJPOXFNFBOBOZWBMJEBUJPOFYFSDJTF TUJMMPQFSBUJOHJOBWBMJENBOOFS wBMUIPVHIUIJT
UIBUUBLFTQMBDFBGUFSUIFPSJHJOBMWBMJEBUJPOXIFO SFRVJSFTPOMZBOFWBMVBUJPOPGUIFWBMJEBUFETUBUVT
UIFQSPEVDUXBTGJSTUCFJOHJOUSPEVDFE5IFTF PGBTZTUFN*GUIJTMFBETUPVOBDDFQUBCMFSFTVMUT
JODMVEFSFWBMJEBUJPOTEVFUPDIBOHF QFSJPEJDUJNF SFWBMJEBUJPONBZCFFYQFDUFE
CBTFESFWBMJEBUJPO BOEPUIFSDBVTFT8FTUBSU 1FSJPEJDSFWBMJEBUJPOJTBMTPSFRVJSFECZUIF
CZHJWJOHKVTUJGJDBUJPO JF XIZXFNBZOFFEUP 8PSME)FBMUI0SHBOJ[BUJPO 8)0
HVJEFMJOFT
SFWBMJEBUFBTZTUFN
/FYUXFPVUMJOFBTUSBUFHZUP PO(.1
"OOFYPGUIF8)0(.1
QSJPSJUJTFBNPOHUIFTZTUFNTUPCFSFWBMJEBUFEBOE EPDVNFOUFOUJUMFESupplementary Guidelines on
UPEFDJEFPOXIBU IPX BOEXIFOUPSFWBMJEBUF Good Manufacturing Practices: ValidationTUBUFT
8FUIFODPOTJEFSIPXUPJEFOUJGZUIFOFFEGPS i5IFSFTIPVMECFQFSJPEJDSFWBMJEBUJPO BTXFMMBT
SFWBMJEBUJPOBOEUPSFEVDFUIJTOFFE/FYUXF SFWBMJEBUJPOBGUFSDIBOHFTw5IFGBDUUIBU8)0JTBO
EJTDVTTUIFDPTUTBTTPDJBUFEXJUISFWBMJEBUJPO JOUFSOBUJPOBMCPEZ BTPQQPTFEUPBNVMUJOBUJPOBM
BOEGJOBMMZUIFEJGGFSFOUCVTJOFTTNPEFMTGPS POFTVDIBTUIF&6PSUIF1IBSNBDFVUJDBM
BQQSPBDIJOHUIFQSPCMFN5ISPVHIPVUXFBUUFNQU *OTQFDUJPO$POWFOUJPO1IBSNBDFVUJDBM*OTQFDUJPO
UPJOUFHSBUFUIFCVTJOFTTEFDJTJPOTXJUIUIF $PPQFSBUJPO4DIFNF<1*$4>
JOEJDBUFTBEFHSFF
UFDIOJDBMPOFTJOBQSBHNBUJDXBZ$POTJEFSBUJPOT PGXPSMEXJEFDPOTFOTVTFTQFDJBMMZPVUTJEFUIF*$)
JOUIFTFEJTDVTTJPOTEPOPUDPNQSPNJTFQBUJFOU countries.
TBGFUZBOEQSPEVDURVBMJUZUIFTFBSFBHJWFOJOUIF Other regulations introduced in the past such
QIBSNBDFVUJDBMJOEVTUSZGPSFUIJDBM SFHVMBUPSZ BOE BTUIF64'PPEBOE%SVH"ENJOJTUSBUJPOT
CVTJOFTTSFBTPOT JF DPSQPSBUFJNBHFBOEBWPJEJOH $'31BSUBEESFTTJOHFMFDUSPOJDSFDPSETBOE
litigations). FMFDUSPOJDTJHOBUVSFTOFDFTTJUBUFEUIFSFWBMJEBUJPO
*OPVSDPOUFYU XFXJMMVTFUIFUFSNTZTUFN PSBUMFBTUUIFBTTFTTNFOUPGMFHBDZTZTUFNT
JOBCSPBETFOTFUPJODPSQPSBUFUIFQSPDFTT UIF UIBUXFSFJOQMBDFCFGPSFTVDISFHVMBUJPOTXFSF
FRVJQNFOUPOXIJDIJUJTJNQMFNFOUFEBOEUIF JOUSPEVDFE(JWFOUIFDIBOHJOHOBUVSFPG
procedure to follow in executing it. This discussion UFDIOPMPHZ JUJTOPUVOSFBTPOBCMFUPFYQFDUGVSUIFS
JTBQQMJDBCMFUPCPUITFDPOEBSZESVHNBOVGBDUVSJOH DIBOHFTJOSFHVMBUJPOTJOUIFGVUVSFUIBUNBZ
JF NBOVGBDUVSFPGEPTBHFGPSNT
BOEQSJNBSZ DBVTFSFWBMJEBUJPOPGFYJTUJOHTZTUFNT1"5JTBO
NBOVGBDUVSJOH JF BDUJWFQIBSNBDFVUJDBM FYBNQMFPGTVDIBUFDIOPMPHZEFQFOEJOHPOIPX
JOHSFEJFOUT<"1*>NBOVGBDUVSJOH
1"5JTVUJMJ[FE NPOJUPSJOHBOEEBUBDPMMFDUJPO
WTJOUFHSBUJOHJOUPBDPOUSPMMPPQ
UIJTNBZDBVTF
WHY REVALIDATE? QSPDFTTDIBOHFOFDFTTJUBUJOHSFWBMJEBUJPO
*UJTOPUVODPNNPOJOUIFQIBSNBDFVUJDBM "OPUIFSSFBTPOGPSSFWBMJEBUJPOJTUIFDIBOHFJO
JOEVTUSZUPWJFXWBMJEBUJPOBTBPOFUJNFFWFOUUP SFHVMBUPSZFYQFDUBUJPOTFWFOXIFOUIFSFHVMBUJPOT
CFQFSGPSNFEPODF BOEOPUUPCFSFWJTJUFEVOMFTT UIFNTFMWFTEPOPUDIBOHF BTNBOJGFTUFECZDIBOHF
BNBKPSDIBOHFUPBTZTUFNUBLFTQMBDF*OGBDU in focus of regulatory inspections. This is also
WBMJEBUJPOTIPVMECFSFHBSEFEBTBTUBUFUPNBJOUBJO FWJEFOUJOUIFDIBOHFTBOEVQEBUFTUPFYJTUJOH
SBUIFSUIBOBOBDUJWJUZUPQFSGPSN HVJEBODFEPDVNFOUTCZSFHVMBUPSZBVUIPSJUJFT
5IFSFBSFTFWFSBMSFBTPOTGPSSFWBMJEBUJOHBTZTUFN PSUIFJOUSPEVDUJPOPGOFXPOFT"QSPNJOFOU
BGUFSJUIBTCFFOJOPQFSBUJPOGPSTPNFUJNF4PNF FYBNQMFPGUIBUJTBHBJOUIFDIBOHFJOSFHVMBUPSZ
PGUIFTFBSFEFTJSBCMFXIJMFPUIFSTBSFNBOEBUPSZ FYQFDUBUJPOTXJUISFHBSEUPUIFJNQMFNFOUBUJPO
5BCMF*
5IFNPTUDPNQFMMJOHJTUIBUJUJTPGUFO PG1BSUFWFOUIPVHIUIFSFHVMBUJPOJUTFMGEJE
BSFHVMBUPSZSFRVJSFNFOU3FHVMBUPSZBVUIPSJUJFT OPUDIBOHF
5IJTIBTBMTPCFFOUIFDBTFXJUI
NBOEBUFSFWBMJEBUJPOBGUFSBDIBOHFUPBQSPDFTT WBMJEBUJPOBMUPHFUIFSPWFSUIFQBTUUISFFEFDBEFT
JODMVEJOHFRVJQNFOU JOTUSVNFOUT QSPDFEVSFT $IBOHFTJOSFHVMBUPSZFYQFDUBUJPOTDBOBMTP
BOENFUIPET
UIBUNBZJNQBDUQSPEVDURVBMJUZ5IF CFJOGFSSFEGSPNXBSOJOHMFUUFSTPSDPNNFOUT
&VSPQFBO6OJPO &6
HPPENBOVGBDUVSJOHQSBDUJDF GSPNQSFJOTQFDUJPOBTTFTTNFOUTCZSFHVMBUPSZ
(.1
HVJEFMJOFSFGFSTUPQFSJPEJDSFWBMJEBUJPO BVUIPSJUJFT*UTIPVMECFOPUFEUIBUUIF
JSSFTQFDUJWFPGTVDIDIBOHF
$IBQUFSGJWFJO '%"ESBGUHVJEBODFGPSQSPDFTTWBMJEBUJPO
IBT
UIF&6(.1EPDVNFOUFOUJUMFEi1SPEVDUJPOw SFNPWFEUIFFYQMJDJUSFGFSFODFUPSFWBMJEBUJPOUIBU
TUBUFT i1SPDFTTFTBOEQSPDFEVSFTTIPVMEVOEFSHP XBTJOUIFPSJHJOBMHVJEBODFEPDVNFOU

BUBMMUJNFT8IFOSFWBMJEBUJOHTVDITZTUFNT
Table I: Possible drivers for revalidation. DPOTUSBJOUTPOSFTPVSDFTJODMVEJOHUJNF DPTU
Changes in the process BOEFYQFSUJTF JF CVTJOFTTDPOTJEFSBUJPOT
NBLF
Changes in regulations QSJPSJUJ[BUJPOPGSFWBMJEBUJPOJOFWJUBCMF5IF
GJSTUTUFQJTUIVTUPQSJPSJUJ[FBNPOHUIFEJGGFSFOU
Changes in regulatory expectations of current regulations
TZTUFNTOFFEJOHSFWBMJEBUJPO
Changes in industry or company best practices 5IFDPNNPOBQQSPBDIJTUPTUBSUXJUITZTUFNT
Maintaining an updated view of the process, audit readiness VTFEGPSQBSFOUFSBMT GPMMPXFECZUIPTFGPSPUIFS
less critical products such as orals then topicals.
*OTUFBEJUTVHHFTUTBTZTUFNMJGFDZDMFBQQSPBDI )PXFWFS XFCFMJFWFUIBUUIFQSJPSJUJ[BUJPOTIPVME
UIBUJODPSQPSBUFTiDPOUJOVFEQSPDFTTWFSJGJDBUJPO w BMTPJODMVEFBCVTJOFTTBTQFDU JF TZTUFNTUIBU
XIJDIJOFTTFODFJTWBMJEBUJPONBJOUFOBODF QSPEVDFQSPEVDUTUIBUBSFDSJUJDBMGPSUIFTVSWJWBMPG
$IBOHFTUIBUPDDVSJOUIFQSBDUJDFPGWBMJEBUJPOJO UIFCVTJOFTTTVDIBTCMPDLCVTUFSESVHT
1SPEVDJOH
JOEVTUSZNBZDBVTFBGJSNUPSFWBMJEBUFJUTTZTUFNT BQBSFOUFSBMUIBUJTOPUGJOBODJBMMZWJBCMFXIJMF
UPBMJHOJUTFMGXJUIDVSSFOUCFTUQSBDUJDFT5IJTJT JHOPSJOHBMFTTDSJUJDBMZFUQSPGJUBCMFQSPEVDUXJMM
FYFNQMJGJFECZUIFJOUSPEVDUJPOPGUIF*OUFSOBUJPOBM NFBOUIBUFWFOUVBMMZQSPEVDUJPOPGUIFGPSNFSXJMM
4PDJFUZGPS1IBSNBDFVUJDBM&OHJOFFSJOH *41&
OPUCFTVTUBJOBCMFJGUIFCVTJOFTTGBJMT
Baseline Guide 5: Commissioning & Qualification and "MMQSPEVDUTBSFSFRVJSFEUPCFNBOVGBDUVSFE
JUTJNQBDUPOUIFXBZRVBMJGJDBUJPOJTQFSGPSNFE
XJUISPCVTUMZWBMJEBUFEQSPDFTTFT8IFO
4VDIDPODFQUTBTJNQBDUBTTFTTNFOUBOEFOIBODFE SFTPVSDFTBSFMJNJUFE IPXFWFS UIFSFBSFTIPSU
EFTJHOSFWJFXIBWFCFDPNFNPSFGPSNBMJ[FEBOE UFSNBMUFSOBUJWFTUPSFWBMJEBUJPO 5BCMF**
0OF
DPNNPOQMBDFXJUIUIFJOUSPEVDUJPOPGUIJTHVJEF TVDIBMUFSOBUJWFJTUPUFNQPSBSJMZEJTDPOUJOVF
5IJTEPFTOPUOFDFTTBSJMZNFBOSFQFBUJOHUIF QSPEVDUJPOPGPOFQSPEVDUVOUJMSFWBMJEBUJPOPG
XIPMFWBMJEBUJPOCVUQPTTJCMZPOMZQSPEVDJOHUIF PUIFSNPSFGJOBODJBMMZWJBCMFPOFTJTDPNQMFUFE
BGPSFNFOUJPOFEEPDVNFOUTJOSFUSPTQFDUBTBOBJE "TFDPOEBMUFSOBUJWFGPSBNVMUJTJUFGJSNJTUP
for future change control. USBOTGFSQSPEVDUJPOUPBOPUIFSTJUFUIBUEPFTIBWF
5IFJTTVFTPVUMJOFEBCPWFSFMBUFEUPSFHVMBUJPOT UIFSFTPVSDFTGPSWBMJEBUJOHUIFQSPDFTT"UIJSE
BOEJOEVTUSZCFTUQSBDUJDFTDBOCFSFHBSEFEBT BMUFSOBUJWFJTUPUFNQPSBSJMZDPOUJOVFQSPEVDUJPO
FYUFSOBMUPBGJSNJOUIFTFOTFPGCFJOHCFZPOE CVUXJUIBEEJUJPOBMQSPDFEVSBMDPOUSPMTTVDIBT
its control. In contrast there are other causes JOTQFDUJPO BEEJUJPOBMJOQSPDFTTDPOUSPMT
GPSSFWBMJEBUJPOUIBUBSFDPOTJEFSFEJOUFSOBMUPB BOENPSFGSFRVFOUQSPEVDUSFWJFXT5IJTJTTJNJMBS
GJSN"MTP NFSHFSTBOEBDRVJTJUJPOTNBZQSPNQU UPUIFDBTFPGNBOVGBDUVSJOHBDUJWFQIBSNBDFVUJDBM
SFWBMJEBUJPOUPBMJHOEJGGFSFOUTUBOEBSETBOE JOHSFEJFOUT "1*T
GPSJOWFTUJHBUJPOBMQSPEVDUTBT
QSBDUJDFTPGWBMJEBUJPOCFUXFFOUIFPSHBOJ[BUJPOT PVUMJOFECZUIF*$)(.1HVJEFGPS"1* Q7
*O
FTQFDJBMMZJGUIFZBSFHMPCBMDPSQPSBUJPOT5IFSFBSF UIFVOMJLFMZFWFOUUIBUOPOFPGUIFBCPWFPQUJPOT
UIFVTVBMDBVTFTGPSSFWBMJEBUJPOTVDIBTGPMMPXJOH JTBUUBJOBCMF QSPEVDUJPODBOCFDPOUSBDUFEPVU CVU
BDIBOHFUIBUNBZJNQBDUQSPEVDURVBMJUZ BOEUP JOUIJTDBTFWBMJEBUJPOBUUIFDPOUSBDUTJUFNVTUUIFO
NBJOUBJOBWBMJEBUFETUBUVTPGBTZTUFN CFDPOTJEFSFE5IFTFBSFUFNQPSBSZNFBTVSFTBOE
"GVSUIFSNPUJWBUJPOJOBTTFTTJOHUIFWBMJEBUFE BSFOPUNFBOUUPSFQMBDFQFSGPSNJOHUIFSFRVJSFE
TUBUVTPGBTZTUFNJTUIFCFOFGJUTBDDSVFEGSPN SFWBMJEBUJPOFYFSDJTF VOMFTTPGDPVSTFNPWJOH
TVDIBOFOEFBWPVS FTQFDJBMMZGSPNBCVTJOFTT BQSPEVDUUPBOPUIFSTJUF DPOUSBDUJOHJUPVUPS
QFSTQFDUJWF5IFTFJODMVEFNBJOUBJOJOHBDVSSFOU QPTTJCMZMJDFOTJOHJUPSFWFOTFMMJOHUIFJOUFMMFDUVBM
WJFXBOEDPOUSPMPGUIFQSPDFTTBOEFOTVSJOHUIBUJU QSPQFSUZSJHIU *13
UPBOPUIFSNBOVGBDUVSFSJTQBSU
JTBMXBZTJOBOBVEJUSFBEZTUBUF5IJTVQEBUFNBZ PGUIFGJSNTTUSBUFHJDPCKFDUJWFT
NFSFMZJOWPMWFVQEBUJOHUIFEPDVNFOUBUJPOBOE "GUFSQSJPSJUJ[BUJPOPGUIFEJGGFSFOUTZTUFNTUPCF
EPFTOPUOFDFTTBSJMZSFRVJSFBGVMMGMFEHFEUFTUJOH SFWBMJEBUFE QSJPSJUJ[BUJPOXJUIJOBTZTUFN 'JHVSF
SFHJNF
JTBTTFTTFE5IJTNFBOTEFDJEJOHPOXIJDI
GVODUJPOTBOEGFBUVSFTXJUIJOBTZTUFNUPWBMJEBUF
SETTING A REVALIDATION STRATEGY 4VDIEFDJTJPOTXJMMMJLFMZCFEJGGFSFOUGSPNUIPTFPG
0ODFUIFOFFEGPSSFWBMJEBUJPOIBTCFFOJEFOUJGJFE UIFPSJHJOBMWBMJEBUJPO5IFPSJHJOBMWBMJEBUJPONBZ
BTUSBUFHZJTEFUFSNJOFE4QFDJGJDBMMZ UIJTJODMVEFT include health and safety issues addressed as an
XIBU IPX BOEXIFOUPSFWBMJEBUF BEKVODUUPUIFWBMJEBUJPOFYFSDJTFGPSDPOWFOJFODF
BOEDPTUSFBTPOT DPNNJTTJPOJOH
5IJT IPXFWFS
What To Revalidate OFFEOPUCFUIFDBTFJOSFWBMJEBUJPO
$MFBSMZBMMTZTUFNTXJUIQPUFOUJBMJNQBDUPO 5IFSFRVJSFNFOUTUPCFWFSJGJFEGPSBOZTZTUFN
QSPEVDURVBMJUZNVTUCFLFQUJOBWBMJEBUFETUBUF DBOCFDMBTTJGJFEJOUPUXPNBJODBUFHPSJFTCVTJOFTT

Figure 1: Prioritization for revalidation.

Table II: Temporary options for a process
awaiting revalidation. Requirements
Discontinue production until revalidation is ready

Transfer production to another site within the organization capable
of validating the process
Impose additional procedural controls (such as 100% inspection)
Outsource manufacturing
Business Regulatory
SFRVJSFNFOUT FH MJOFDBQBDJUZ

BOESFHVMBUPSZ
SFRVJSFNFOUT 'JHVSF
5IFMBUUFSJOUVSOBSF
GVSUIFSDMBTTJGJFEJOUP(91SFRVJSFNFOUTBOE
PUIFSTJODMVEJOHFOWJSPONFOUBM IFBMUI BOETBGFUZ
&)4
BOETUBUVUPSZ8FCFMJFWFUIBUJOUIFDBTF
PGSFWBMJEBUJPO POMZ(91SFRVJSFNFOUTBSFUPCF GXP Other
WFSJGJFE5IJTEPFTOPUNFBOUIBUPUIFSSFHVMBUPSZ (e.g., EHS, Statutory)
BOECVTJOFTTSFRVJSFNFOUTBSFVOJNQPSUBOU CVU
UIBUUIFZTIPVMECFWFSJGJFEPVUTJEFUIFTDPQFPGUIF
SFWBMJEBUJPO Prioritization within a system
5IF(91SFRVJSFNFOUTUPCFWFSJGJFEXJMMCF
EFUFSNJOFECZUIFBTTFTTNFOUPGUIFJSJNQBDUPO EHS = Environmental, Health, and Safety
QSPEVDURVBMJUZ*UJTCFDPNJOHDPNNPOQMBDFJO
WBMJEBUJPOUPFNQMPZBSJTLNBOBHFNFOUQSPDFTT
GPSFYBNQMFBTPVUMJOFEJO*$)Q9
XIFSFUIF FYBNQMFTBSFMJOFTQFFEGPSBQBDLBHJOHMJOFPS
QIJMPTPQIZJTUIBUUIFFGGPSUTIPVMECFDPNNFOTVSBUF speed of a depyrogenation tunnel that is known to
with the potential risk. This is also in line with PQFSBUFBUBTQFFENVDITMPXFSUIBOJUTNBYJNVN
UIF'%"SJTLCBTFEBQQSPBDIUP(.1
0OFPG UPHVBSBOUFFQFSGPSNBODF"OPUIFSFYBNQMFJTGPS
UIFDPNNPONFUIPETVTFEGPSSJTLJEFOUJGJDBUJPO BVUPDMBWFTUIBUOPSNBMMZJODMVEFTFWFSBMUZQFTPG
BTTFTTNFOU BOENJUJHBUJPOQMBOOJOHJTUIFGBJMVSF TUFSJMJ[BUJPODZDMFT5IFSFJTOPOFFEUPWBMJEBUF
NPEFFGGFDUBOBMZTJT '.&"
NFUIPE UIFQFSGPSNBODFPGBMMDZDMFTJGUIFZBSFOPUVTFE
5IFWBMJEBUJPOSFQPSUNVTUDMFBSMZTQFDJGZUIF
How To Revalidate QBSBNFUFSTGPSXIJDIUIFTZTUFNJTWBMJEBUFEJODBTF
/PXXFUVSOPVSBUUFOUJPOUPUIFFYUFOUSFWBMJEBUJPO UIFFRVJQNFOUJTOFFEFEJOUIFGVUVSFGPSBEJGGFSFOU
JTUPCFQFSGPSNFE JF TFUUJOHUIFBDDFQUBODFDSJUFSJB process.
CZXIJDIUPKVEHFUIFQSPDFTT
*OBOFXQSPKFDU BOE 5IFQVSQPTFPGSFWBMJEBUJPOJTUPWFSJGZQSPDFTT
XIFOPSEFSJOHOFXFRVJQNFOU UIFGVMMFYUFOUPGJUT SFRVJSFNFOUTSBUIFSUIBOPUIFSFOHJOFFSJOH
VTBHFJTOPUBMXBZTLOPXO4QFDJGJDBUJPOTBSFUIVTTFU SFRVJSFNFOUTTVDIBTUISPVHIQVUPSDBQBDJUZ
CPUIGPSDVSSFOUBOEFYQFDUFEGVUVSFVTBHF'BDUPSZ .FFUJOHUIFTFPUIFSSFRVJSFNFOUTEPFTOPU
BDDFQUBODFUFTUJOH '"5
TJUFBDDFQUBODFUFTUJOH HVBSBOUFFNFFUJOHQSPEVDUTQFDJGJDBUJPOT5IF
4"5
BOEWBMJEBUJPOPGUFOUFTUBHBJOTUTQFDJGJDBUJPOT QSPCMFNPGMPTTJOUISPVHIQVUXJMMTUJMMOFFEUPCF
UIBUDPSSFTQPOEUPUIFQFSGPSNBODFMJNJUTPGUIF BEESFTTFEBMCFJUPOBNPSFTUSBUFHJDMFWFMUIBOUIBU
FRVJQNFOU0OFPGUIFQVSQPTFTPGUIBUJTUPFOTVSF PGSFWBMJEBUJPO TVDIBTTJUFDBQBDJUZFYQBOTJPOPS
UIBUDPOUSBDUVBMPCMJHBUJPOTPOUIFTVQQMJFSBSFNFU NBKPSFRVJQNFOUPWFSIBVM
*OSFWBMJEBUJPO IPXFWFS UFTUJOHBHBJOTUBNVDINPSF
OBSSPXPQFSBUJOHSBOHFNBZCFBQQSPQSJBUF When To Revalidate
3FWBMJEBUJPOTIPVMECFBHBJOTUUIFDVSSFOU "DDPSEJOHUPBMM(.1SFHVMBUJPOT BOZDIBOHF
QSPDFTTSFRVJSFNFOUT*GUIPTFEPOPUGPSNBMMZ XJUIQPUFOUJBMJNQBDUPOQSPEVDURVBMJUZNVTUCF
FYJTU BiSFBMJTUJDwVTFSSFRVJSFNFOUTQFDJGJDBUJPO BTTFTTFEUISPVHIDIBOHFDPOUSPMSFWBMJEBUJPONBZ
634
TIPVMECFEFWFMPQFEJOSFUSPTQFDUUPQSPWJEF CFSFRVJSFEBTBQPTTJCMFPVUDPNF"EEJUJPOBMMZ BT
BDDFQUBODFTQFDJGJDBUJPOT&YBNQMFTPGTVDIBDBTF PVUMJOFEJOUIF&6(.1BOEUIF8)0HVJEFMJOFT
JODMVEFBNJYFSUIBUJTPQFSBUFEBUDFSUBJOTQFFET UIJTBTTFTTNFOUTIPVMECFDBSSJFEPVUQFSJPEJDBMMZ
UIBUBSFMFTTUIBOJUTNBYJNVNQPTTJCMF5IFSF 8FMPPLBUUIPTFJTTVFTIFSFBOEUSZUPQVUUIFN
JTOPQPJOUJORVBMJGZJOHUIFNBYJNVN XIJDI JOBVOJGJFEGSBNFXPSL8FDMBTTJGZUIFOFFEGPS
JTOPUFNQMPZFEJOUIFQSPDFTTBOENBZQPTTJCMZ SFWBMJEBUJPOJOUPFWFOUCBTFEBOEUJNFCBTFE
CFVOBDIJFWBCMFBTXJUIPMEFRVJQNFOU0UIFS 'JHVSF


Figure 2: When to revalidate. DIBOHFJTJNQMFNFOUFE5IJTXJMMFOTVSFOPUPOMZ

Revalidation UIBUUIFSFTVMUJOHTZTUFNBGUFSDIBOHFJTDPNQMJBOU
CVUBMTPUIBUPUIFSQSPDFTTFTBSFOPUBGGFDUFEXIJMF
UIFDIBOHFJTUBLJOHQMBDF"OFYBNQMFPGUIJTJT
XIFONPEJGJDBUJPOTBSFNBEFUPBIBMMPSBCVJMEJOH
Time based Event based UIBUIPVTFTPUIFSTZTUFNT PSUPVUJMJUZMJOFTUIBU
GFFEPUIFSQSPDFTTFT"MMQMBOOFEBOEVOQMBOOFE
DIBOHFTSFRVJSFVQEBUJOHEPDVNFOUBUJPOBOE
SFDPSETQPTUNPEJGJDBUJPO
Calendar time Operating time Inadvertent Predecided Time-based Revalidation.5JNFCBTFE
(unplanned) (planned)
SFWBMJEBUJPO BTJTFWJEFOUGSPNUIFUFSN JT
QFSGPSNFEQFSJPEJDBMMZBOEJTNPTUDPNNPOJO
Event-based Revalidation.&WFOUCBTFE BTFQUJDQSPDFTTJOHBOETUFSJMJ[BUJPO*UJTSFRVJSFE
SFWBMJEBUJPOFTTFOUJBMMZSFGFSTUPSFWBMJEBUJPO CZUIF8)0HVJEFMJOFTPO(.15IFUJNJOHDBO
GPMMPXJOHBDIBOHF5IJTDIBOHFDBOCFFJUIFS CFCBTFEPOPQFSBUJOHUJNFPSDBMFOEBSUJNF
JOBEWFSUFOUPSQSFEFUFSNJOFE5IFGPSNFSNFBOT 5IFDIPJDFPGXIJDIBQQSPBDIUPUBLFJTCBTFEPO
UIBUUIFDIBOHFIBTUPCFQFSGPSNFEJOPSEFSUPCF UIFSJTLUPQSPEVDURVBMJUZ8IFOBQQSPQSJBUF
BCMFUPLFFQQSPEVDJOHQSPEVDUXJUIUIFSFRVJSFE FDPOPNJDGBDUPSTBSFBMTPDPOTJEFSFE"DBMFOEBS
TQFDJGJDBUJPOT'PSFYBNQMF UIJTDBOPDDVSXIFOB UJNFCBTFEBQQSPBDIJTPGUFOUBLFOCFDBVTFJUJT
QBSUXJUIJNQBDUPOQSPEVDURVBMJUZNBMGVODUJPOT FBTJFSUPQMBOBOEJNQMFNFOUIPXFWFS UIJTJT
BOEIBTUPCFSFQMBDFE5IJTLJOEPGDIBOHFJT OPUOFDFTTBSJMZBMXBZTUIFCFTUDIPJDFGSPNBO
usually unplanned. FDPOPNJDQPJOUPGWJFX$POTJEFSUIFDBTFPGB
"OPUIFSFYBNQMFJTXIFOBDIBOHFJOSFHVMBUJPO TZTUFNUIBUJTSBSFMZVTFEEVFUPMPXEFNBOEPO
DBVTFTBQBSUUPCFDPNFVOTVJUBCMFGPSQSPEVDUJPO JUTQSPEVDU*UNBZOPUCFFDPOPNJDBMMZTPVOE
FWFOJGJUJTGVODUJPOJOHDPSSFDUMZQPTTJCMZEVFUP UPWBMJEBUFJUQFSJPEJDBMMZJOTQJUFPGUIFMPX
DIBOHFTJONBUFSJBMDFSUJGJDBUJPOSFRVJSFNFOUT VUJMJ[BUJPO BTUIFSFJTQPUFOUJBMGPSJUOPUCFJOHVTFE
*OTVDIBDBTF UIFSFQMBDFNFOUXJMMIBWFUPCF CFUXFFOUXPDPOTFDVUJWFWBMJEBUJPOT0OUIFPUIFS
SFWBMJEBUFEJGUIFPSJHJOBMQBSUXBTEFFNFE(.1 FYUSFNF UIFSFJTBTZTUFNUIBUJTJODPOUJOVBMVTBHF
DSJUJDBM"GJSNEPFTOPUIBWFUIFDIPJDFPGOPU QPTTJCMZPOTIJGUCBTJTGPSTVDIBTZTUFN UBLJOH
QFSGPSNJOHUIFSFWBMJEBUJPOIFODF XFUFSNJU BQVSFMZDBMFOEBSUJNFCBTFEBQQSPBDIDBOCFUPP
JOBEWFSUFOU SJTLZUPUIFQSPEVDURVBMJUZ*GBUSFWBMJEBUJPOUIF
1SFEFUFSNJOFEDIBOHFTBSFQMBOOFEBIFBEPG TZTUFNJTGPVOEUPCFJOTJHOJGJDBOUEFWJBUJPO UIJT
UJNF5IPTFDBOCFGVSUIFSDMBTTJGJFEBDDPSEJOH XPVMEKFPQBSEJ[FUIFFOUJSFQSPEVDUUIBUIBTCFFO
UPUIFQVSQPTFJOUPSFBDUJWFBOEQSPBDUJWF5IF QSPEVDFETJODFUIFMBTUTVDDFTTGVMWBMJEBUJPO*O
GPSNFSBSFDIBOHFTJOSFBDUJPOUPBSFDVSSJOH TVDIBDBTF SFWBMJEBUJPOCBTFEPOUIFOVNCFSPG
QSPCMFN XIJDI XIJMFOPUDBVTJOHTUPQQBHFPSOPO CBUDIFTQSPEVDFENBZCFTBGFSCPUIUPQSPEVDU
DPOGPSNJUZ NBZDBVTFSFQFBUFEQSPEVDUQSPCMFNT RVBMJUZBOEFDPOPNJDBMMZ5IVTJOQSBDUJDBMUFSNT
1SPBDUJWFDIBOHFTBSFPOFTXIPTFQVSQPTFJTUP BOEUPSFEVDFUIFSJTLUPQSPEVDURVBMJUZ UIFTIPSUFS
JNQSPWFUIFQSPDFTTXJUIPVUJUDVSSFOUMZTVGGFSJOH PGCPUIUJNFTTIPVMECFUBLFO
BOZQSPCMFNT'PSFYBNQMF UIJTDBOOBSSPX 5IFJEFBPGPQFSBUJOHUJNFCBTFEBDUJPOTJT
QSPEVDUWBSJBCJMJUZJOTQJUFPGCFJOHBMSFBEZJO DPNNPOJOQSFWFOUJWFNBJOUFOBODFQSPHSBNT
TQFD PSGPSUFDIOJDBMJNQSPWFNFOUMJLFSFQMBDJOHB XIFSFCZEJGGFSFOUQBSUTBSFSFQMBDFEBGUFSDFSUBJO
NBDIJOFQBSUXJUIBNPSFFGGJDJFOUPOF*UDBOBMTP IPVSTJOPQFSBUJPO*OWBMJEBUJPOUIJTDPSSFTQPOET
CFGPSQVSFMZFDPOPNJDSFBTPOTTVDIBTJODSFBTJOH UPUIFOVNCFSPGCBUDIFTQSPEVDFE
UISPVHIQVUPSSFEVDJOHFOFSHZDPOTVNQUJPO The essence of this section is that the choice
4VDIQSPBDUJWFDIBOHFTDBOGBMMVOEFSQSPDFTT PGXIFOUPSFWBMJEBUF XIFUIFSJUJTFWFOUCBTFE
PQUJNJ[BUJPO5IPTFXPVMEOBUVSBMMZUBLFMPXFS QMBOOFEPSVOQMBOOFEDIBOHF
PSUJNFCBTFE
priority than unplanned changes whose purpose is PQFSBUJOHPSDBMFOEBSUJNF
TIPVMEVMUJNBUFMZCF
UPLFFQQSPEVDUJPOHPJOHBOENBJOUBJOJOHQSPEVDU CBTFEPOQPUFOUJBMSJTLUPUIFQSPEVDU
RVBMJUZ 5IFDPODFQUTPGUJNFCBTFEBOEFWFOUCBTFE
*OFJUIFSDBTF XIFUIFSUIFDIBOHFJTQMBOOFEPS BDUJPOTDPNFGSPNUIFGJFMEPGQSPHSBNNBCMF
VOQMBOOFE BSJHPSPVTDIBOHFDPOUSPMQSPDFEVSF DPOUSPM XIFSFUJNFCBTFESFGFSTUPUJNFSTXIJMF
TIPVMECFJOQMBDF'PSUIFQMBOOFEDIBOHFT UIJT FWFOUCBTFESFGFSTUPDPVOUFST BDUJWBUJPOPGMJNJU
TIPVMETUJQVMBUFUIFBTTFTTNFOUPGUIFDIBOHFBOE TXJUDIFT PSPUIFSDPOEJUJPOT4USJDUMZTQFBLJOH POF
JUTJNQBDUPOQSPEVDURVBMJUZBOEQSPDFTTJOUFHSJUZ DBODPOTJEFSUJNFCBTFEBDUJPOTBTFWFOUCBTFEPOFT
JOBEEJUJPOUPIFBMUIBOETBGFUZJTTVFTCFGPSFUIF XIFSFCZUIFFMBQTJOHPGUIFUJNFQFSJPEJTJUTFMGUIF

FWFOU)PXFWFS UIFDPOWFOUJPOIBTCFFOUPUSFBU Figure 3: Some means for detecting and reduc-

UIFNBTEJTUJODU ing the need for revalidation.
Change
HOW TO IDENTIFY AND REDUCE THE
NEED FOR REVALIDATION
6MUJNBUFMZPOFDBOBSHVFUIBUUIFOFFEGPS
SFWBMJEBUJPOBSJTFTEVFUPUIFPDDVSSFODFPGB Detection Reduction
DIBOHF5IJTDIBOHFNBZCFLOPXOBOEEFDJEFE
XIFUIFSQMBOOFEPSVOQMBOOFE"MUFSOBUJWFMZ JU
DBOPDDVSXJUIPVUCFJOHEFUFDUFE XIJDIUJNF
CBTFESFWBMJEBUJPOBUUFNQUTUPNJUJHBUF
BT
SPC PQR PM Calibration
XBJUJOHGPSUIFDIBOHFUPNBOJGFTUJUTFMGJOBOPVU
PGTQFDJGJDBUJPOQSPEVDUJTVOBDDFQUBCMF)FODFUP Detection of gradually occuring changes Reduction of the need for a change
UBLFBNPSFTUSVDUVSFEBQQSPBDI UIFSFTIPVMECF SPC = Statistical Process Control PM = Planned Maintenance

PQR = Product Quality Review
NFDIBOJTNTUPEFUFDUBOEJGQPTTJCMFQSFEJDUUIJT
DIBOHFBTTPPOBTQPTTJCMF TFF'JHVSF
4VDI
NFDIBOJTNTBMSFBEZFYJTUJOQSBDUJDFBMUIPVHIOPU SFRVJSFEGVODUJPOBMJUZBTTUBOEBSE5IJTJODMVEFT
BMXBZTUJFEUPWBMJEBUJPO data gathering and processing such as calculating
NFBOT TUBOEBSEEFWJBUJPOT BOEDPOUSPMMJNJUT
Statistical Process Control &YBNQMFTPGTVDITZTUFNTJODMVEFNPEFSOUBCMFU
4UBUJTUJDBMQSPDFTTDPOUSPM 41$
BMMPXTUIF QSFTTFTBOEUBCMFUBVUPUFTUFST8IFOBQQMZJOH41$
DPOUJOVPVTNPOJUPSJOHPGUIFQSPDFTTBOEDBO POFTIPVMECFBXBSFPGJUTVOEFSMZJOHBTTVNQUJPOT
EFUFDUHSBEVBMTIJGUTGSPNJUTDVSSFOU JEFBMMZ BOE DPOTFRVFOUMZ JUTMJNJUBUJPOT
DPOUSPMMFE
TUBUF5IFCBTJTPG41$JTUPNBJOUBJO
UIFQSPDFTTJOBTUBUFPGiTUBUJTUJDBMDPOUSPM w Product Quality Review
NFBOJOHUIBUJUTPOMZWBSJBCJMJUZJTEVFUPiDIBODF "OPUIFSNFDIBOJTNUIBUJTDVSSFOUMZVTFEJTUIF
DBVTFTwBOEEPFTOPUJODMVEFBOZiBTTJHOBCMF QSPEVDURVBMJUZSFWJFX XIJDIIBTUPCFDPOEVDUFE
DBVTFTw
"TTVNJOHUIBUBQSPDFTTJTJOTUBUJTUJDBM QFSJPEJDBMMZBTNBOEBUFECZUIF(.1T"TQBSU
DPOUSPMBUJODFQUJPO XIFOBO41$DIBSUEFUFDUTBO PGUIJTFYFSDJTF UIF&6(.1TUBUFT i"SFWJFXPG
PVUPGDPOUSPMDPOEJUJPO UIJTTJHOBMTUIBUBDIBOHF DSJUJDBMJOQSPDFTTDPOUSPMTBOEGJOJTIFEQSPEVDU
IBTPDDVSSFEBOEIBTNBOJGFTUFEJOBOJODSFBTFJO SFTVMUTwTIPVMEUBLFQMBDF JOBEEJUJPOUPBIPTU
QSPDFTTWBSJBCJMJUZ4VDIBTJUVBUJPOXPVMEQSPNQU PGPUIFSSFWJFXT
4VDISFWJFXTXPVMEEFUFDU
JOWFTUJHBUJPOTUIBUNBZSFWFBMQSPCMFNTXJUITPNF BOZNBKPSEFWJBUJPOTJOUIFQSPDFTTBOEXPVME
QBSUPGUIFFRVJQNFOUPSMJOFPOXIJDIUIFQSPDFTT IFODFUSJHHFSBOJOWFTUJHBUJPOUIBUNBZMFBE
JTSVOOJOH*GTVDIBQBSUOFFETSFQMBDFNFOUBOE UPSFWBMJEBUJPOPGPOFPSNPSFPGUIFTZTUFNT
JTEFFNFEUPCFDSJUJDBMUPQSPEVDURVBMJUZ UIFO BTTPDJBUFEXJUIUIFEFWJBUJPO4VDIBQSPEVDU
UIJTNBZMFBEUPSFWBMJEBUJOHUIFQSPDFTTBGUFS RVBMJUZSFWJFXBMSFBEZUBLFTQMBDFPOSFHVMBSCBTJT
SFQMBDFNFOU*GUIFSFQMBDFNFOUJTPGBiMJLFGPS and will not constitute any additional cost. The
MJLFwOBUVSF WBMJEBUJPONBZOPUCFOFDFTTBSZ0O usefulness of this exercise in detecting the need
UIFPUIFSIBOE JGUIFJOWFTUJHBUJPOSFWFBMTUIF GPSSFWBMJEBUJPOEFQFOETPOUIFGSFRVFODZXJUI
SFDVSSFODFPGUIF41$USFOETBGUFSSFQMBDFNFOU XIJDIJUJTDPOEVDUFE"TXJUIUIFDBTFPGQFSJPEJD
PGUIFSFMFWBOUFRVJQNFOUQBSUT UIFOBNBKPS SFWBMJEBUJPO SJTLUPUIFQSPEVDUJTBNBKPSGBDUPS
QSPDFTTDIBOHFNBZCFOFFEFEBTJOEJDBUFEJOUIF JOUIFDIPJDFPGUIFSFWJFXQFSJPE1SPEVDURVBMJUZ
EJTDVTTJPOPGDIBOHFDPOUSPMXJUISFWBMJEBUJPOBT SFWJFXTBSFDPNNPOMZDPOEVDUFEBOOVBMMZ
BIJHIMZMJLFMZPVUDPNF*UTIPVMECFOPUFEUIBU
JOUIFTUBUJTUJDBMTFOTF BOPVUPGDPOUSPMDPOEJUJPO Preventive Maintenance
EPFTOPUOFDFTTBSJMZNFBOBOPVUPGTQFDJGJDBUJPO 5PSFEVDFUIFOFFEGPSSFWBMJEBUJPO TFF'JHVSF
QSPEVDU)PXFWFS TVDIBDPOEJUJPOJGOPU
BGJSNTIPVMEFOTVSFUIFUJNFMZFYFDVUJPOPG
BEESFTTFENBZMFBEUPBOPODPOGPSNJUZ5IVT41$ FGGFDUJWFQSFWFOUJWFNBJOUFOBODFQMBOT5IJTXJMM
TFSWFTBTBOFBSMZXBSOJOHBHBJOTUOPODPOGPSNJUZ IFMQLFFQFRVJQNFOUBUBIJHIMFWFMPGQFSGPSNBODF
7BSJBUJPOTPG41$TVDIBTNVMUJWBSJBUF41$ .41$
SFEVDFWBSJBCJMJUZBOEQFSGPSNBODFQSPCMFNT

BSFNPSFEJTDSJNJOBUJOHJOEJBHOPTJTBOEIFODF BOENJOJNJ[FUIFOFFEGPSSFWBMJEBUJPO"OPUIFS
BNPSFQPXFSGVMUPPM JNQPSUBOUGBDUPSUIBUDPOUSJCVUFTUPSFEVDJOHUIJT
41$JTBDPOWFOJFOUUPPMBOEDBOCFJNQMFNFOUFE OFFEJTTUSJDUMZFYFDVUJOHUIFDBMJCSBUJPOQMBO*U
BUNJOJNBMDPTU.BOZNPEFSOTZTUFNTJODMVEFUIF JTUIFNFBTVSJOHJOTUSVNFOUTBOEEFWJDFTUIBU

EFUFDUBOZEFWJBUJPOTGSPNTQFDJGJDBUJPOT5IVT UIFDPTUJODVSSFEEVFUPUBLJOHUIFFRVJQNFOUPVU
UIFSFMJBCJMJUZPGUIFEFDJTJPOTCBTFEPOUIFJS of production. There are other costs associated with
results including the potentially costly decision to this issue.
SFWBMJEBUFBTZTUFNXJMMEFQFOEPOUIFBDDVSBDZPG *GWBMJEBUJPOEPDVNFOUTBSFOPUBWBJMBCMF
UIPTFNFBTVSFNFOUT SFWBMJEBUJPONBZOFFEUPCFDPOEVDUFE5IJTDBO
CFBMFOHUIZFYFSDJTFEFQFOEJOHPOUIFOBUVSF
COSTS ASSOCIATED WITH PGUIFNJTTJOHEPDVNFOUT.BKPSFGGPSUNBZCF
REVALIDATION SFRVJSFE JODMVEJOHHFOFSBUJOHB634&WFOJGBMM
8FOPXGPDVTPOUIFCVTJOFTTBTQFDUTPGSFWBMJEBUJPO UIFEPDVNFOUTBSFBWBJMBCMF OFXUFTUJOHNBZCF
JODMVEJOHDPTUBOEUIFBMUFSOBUJWFBTTPDJBUFECVTJOFTT SFRVJSFEUPDPNQMZXJUIDVSSFOUSFHVMBUJPOTPS
EFDJTJPOT8FBUUFNQUUPSBUJPOBMJ[FUIFSFWBMJEBUJPO JOEVTUSZFYQFDUBUJPOT$PNQVUFSTZTUFNTBOE
process in the wide sense including aspects of its TPVSDFDPEFBWBJMBCJMJUZXBTOPUBQBTUSFRVJSFNFOU
NBOBHFNFOUQSPDFTT &RVJQNFOUNBZIBWFQSPEVDUDPOUBDUQBSUTPS
5IFNPTUDPTUMZQIBTFPGSFWBMJEBUJPOJTXIFO MVCSJDBOUTUIBUXFSFQSFWJPVTMZBDDFQUBCMFCVUBSF
QFSGPSNJOHJUGPSUIFGJSTUUJNF5IJTJTFTQFDJBMMZTP OPUDVSSFOUMZBQQSPWFE0CUBJOJOHTVDIEBUBDBO
GPSTZTUFNTUIBUIBWFCFFOJOPQFSBUJPOGPSBMPOHUJNF CFBDIBMMFOHF4VQQMJFSTNBZOPUCFJOCVTJOFTT
BTUIFSFJTQPUFOUJBMSJTLPGESJGUJOHGSPNUIFPSJHJOBM "OFRVJQNFOUNPEFMNBZCFEJTDPOUJOVFE
WBMJEBUFETUBUVT5IFIJHIDPTUPGUIFGJSTUSFWBMJEBUJPO BOEBTTPDJBUFEEPDVNFOUBUJPOJTOPUBWBJMBCMF
GPSTVDITZTUFNTJTEVFUPUIFGPMMPXJOHNBJOSFBTPOT &RVJQNFOUNBZTUJMMCFBWBJMBCMF CVUDFSUBJO
4PNFBTQFDUTPGUIFWBMJEBUJPOTUBUVTNBZFJUIFSCF JOGPSNBUJPONBZOPUCFBWBJMBCMFPSNBZOPUCF
OPULOPXOPSDBOOPUCFWFSJGJFE BOEIFODFUIFSFJTB OPSNBMMZHFOFSBUFECZUIFWFOEPS5IJTJTOPU
QPUFOUJBMMZMBSHFDPTUJOWPMWFEJOUIFBTTFTTNFOUPGUIJT BOVODPNNPODBTFXJUITVQQMJFSTGPSXIPNUIF
TUBUVT5IFDVNVMBUJWFFGGFDUPGNJOPSDIBOHFTUPUIF QIBSNBDFVUJDBMNBSLFUSFQSFTFOUTBTNBMMTIBSF
TZTUFNTPJODSFBTFTUIFSJTLPGJUCFDPNJOHTVGGJDJFOUMZ PGUIFJSPWFSBMMCVTJOFTT"HBJOTVDINJTTJOH
EFWJBUFEGSPNJUTJOJUJBMWBMJEBUFETUBUFUIBUUIFDPTUPG JOGPSNBUJPOXJMMIBWFUPCFHFOFSBUFECZUIF
CSJOHJOHJUCBDLJOUPUIJTTUBUFXJMMCFIJHI5IFTFIJHI QIBSNBDFVUJDBMGJSNBOEDBOCFWFSZDPTUMZ TVDIBT
DPTUTXJMMCFEVFUPDBQJUBMJOWFTUNFOUJOUIFTZTUFNJO UIFDBTFXJUIQPTJUJWFNBUFSJBMJEFOUJGJDBUJPO
UIFGPSNPGNBKPSSFQMBDFNFOUPSNBJOUFOBODFPGTPNF *OBEEJUJPO UIFSFJTUIFDPTUPGBDUVBMMZ
PGJUTDPNQPOFOUT)PXFWFS BGUFSTVDIBOPWFSIBVM B QFSGPSNJOHUIFSFWBMJEBUJPOUFTUTFWFOJGBMM
NBKPSDPTUMZWBMJEBUJPOFYFSDJTFXPVMECFOFFEFE TFF UIFOFDFTTBSZJOGPSNBUJPOBOEEPDVNFOUT
'JHVSF
BSFBWBJMBCMF5IJTJODMVEFTUIFUJNF MBCPS
JOTUSVNFOUT BOEUIFQSPEVDUPSQMBDFCPJG
Cost Of Assessment Of The Validation SFRVJSFE'JOBMMZ BTJOUIFDBTFPGBTTFTTNFOU
Status DPTUT UIFSFJTUIFDPTUPGMPTUPQQPSUVOJUZ JF
5IFNBKPSDPOUSJCVUJPOUPUIFBTTFTTNFOUDPTUJTUIF UIFUJNFUIBUUIFFRVJQNFOUJTPVUPGVTFBOEUIF
EJGGJDVMUZJOMPDBUJOHUIFOFDFTTBSZJOGPSNBUJPOBCPVU QFSTPOOFMBSFUBLFOPGGUIFJSSFHVMBSKPCT

BTZTUFN FTQFDJBMMZBCPVUUIFFRVJQNFOUVTFEJO $POTJEFSBUJPOTPGUIFDPTUTBOEBTTPDJBUFE
UIFQSPDFTT5IJTNBZCFEVFUPFRVJQNFOUCFJOHTP BDUJWJUJFTXJMMEFUFSNJOFUIFGVUVSFWBMJEBUJPO
PMEUIBUJUTEPDVNFOUTBSFNJTQMBDFEPSMPTU4VDI QSPHSBNBOEUIFMFWFMPGSFWBMJEBUJPOSFRVJSFE'PS
JNQPSUBOUEPDVNFOUTJODMVEFWBMJEBUJPOGJMFTXJUI FYBNQMF JGJOQSPDFTTEBUBBOEUIFSFHVMBSQSPEVDU
TVQQPSUJOHDFSUJGJDBUFTBOESFQPSUTBOENBDIJOF RVBMJUZSFWJFXTJOEJDBUFBSFDVSSJOHQSPCMFNGSPN
PQFSBUJPOBOENBJOUFOBODFNBOVBMT%PDVNFOUTGPS BHJWFOQJFDFPGFRVJQNFOUBOEJOBEEJUJPOUIJT
FRVJQNFOUSFMPDBUFEGSPNBOPUIFSTJUFNBZIBWFCFFO JTDPVQMFEXJUINJTTJOHPSVOPCUBJOBCMFEBUB
MPTUJOUSBOTJU%PDVNFOUTGPSUSBOTGFSSFEFRVJQNFOU UPWFSJGZJUTQFSGPSNBODF UIFOJUNJHIUCFNPSF
NBZCFBWBJMBCMFCVUJOBEJGGFSFOUMBOHVBHF5IFTF DPTUFGGFDUJWFUPSFQMBDFUIFFRVJQNFOU5IJTXJMM
TDFOBSJPTBSFOPUVOSFBMJTUJDJOMJHIUPGUIFNFSHFST TBWFUIFUJNFBOEDPTUTJOWPMWFEJOBTTFTTJOH
BOEBDRVJTJUJPOT BOEXJUIUIFTVCTFRVFOUDMPTFEPXO UIFTUBUFPGUIJTFRVJQNFOUBOECSJOHJOHJUVQ
PSEPXOTJ[JOHPGGBDJMJUJFT5IFUJNFUBLFOUPMPDBUF UPUIFDVSSFOUWBMJEBUJPOTUBOEBSET*OBXJEFS
EPDVNFOUTBOEJOGPSNBUJPOJOEJSFDUMZUSBOTMBUFTJOUP DPOUFYU UIJTXPVMEBQQMZUPBMMUIFTZTUFNTPO
GJOBODJBMDPTU*OBEEJUJPO UIFSFBSFUIFEJSFDUDPTUT TJUFOFFEJOHWBMJEBUJPOBOEDBOIFMQUPOBSSPX
BTTPDJBUFEXJUIUIFWBMJEBUJPOQFSTPOOFMJOWPMWFEJO EPXOUIFQSJPSJUJ[BUJPOQSPDFTTNFOUJPOFEFBSMJFS
UIJTBTTFTTNFOU .PSFJNQPSUBOUMZ JUXJMMIFMQSFEVDFUIFSJTLTCZ
SFQMBDJOHQSPCMFNBUJDTZTUFNTXJUIOFXFRVJQNFOU
Cost Of Performing The Revalidation UIBUJTFYQFDUFEUPQSPWJEFTVQFSJPSQFSGPSNBODF
5IJTJTUIFDPTUPGBDUVBMMZQFSGPSNJOHUIFUFTUT BOEDPNQMJBODFBOECFNPSFFDPOPNJDJOUIF
SFDPSEJOH BOESFQPSUJOHDPODMVTJPOTJOBEEJUJPOUP MPOHFSUFSN

THE BUSINESS MODEL PGFNQMPZJOHBDPOUSBDUPSJTUIFMPTTPGLOPXMFEHF

0ODFUIFTDPQFPGUIFSFWBMJEBUJPOFYFSDJTFJTEFDJEFE PGUIFQSPDFTTPODFUIFZIBWFMFGU"UUFNQUTTIPVME
UIFCVTJOFTTBQQSPBDIUPFYFDVUJPOJTEFUFSNJOFE CFNBEFUPDBQUVSFBTNVDIPGUIJTLOPXMFEHFBT
5IJTQMBOXJMMJODMVEFUIFCVEHFU UBSHFUT BOE QPTTJCMFXIFUIFSJOXSJUJOHPSJOQFSTPOCZUIF
NJMFTUPOFT8FBSFTUJMMJOUIFDPOUFYUPGUIFGJSTU JOIPVTFTUBGG6OGPSUVOBUFMZ FYQFSJFODFJOEJDBUFT
SFWBMJEBUJPOBTJUJTMJLFMZUPCFUIFNPTUSFTPVSDF XJEFWBSJBCJMJUZJOUIFDPTUBOEDPNNJUNFOUPG
EFNBOEJOH*UJTBMTPUIFNPTUCFOFGJDJBMTJODF WBMJEBUJPODPOUSBDUPST
TVCTFRVFOUSFWBMJEBUJPOTXJMMCFNFSFMZNBJOUBJOJOH 5IFQVSQPTFPGUIJTTFDUJPOJTOPUUPHJWFUIFCFTU
UIFOFXTUBUVTBDIJFWFEUISPVHIDIBOHFDPOUSPM BQQSPBDIPSUPQSPWJEFBPOFTJ[FGJUTBMMTPMVUJPO
5IFSFBSFUXPNBJOCVTJOFTTPQUJPOTGPS CVUSBUIFSUPQSFTFOUEJGGFSFOUPQUJPOT5IFGJOBM
QFSGPSNJOHUIFGJSTUBOENBKPSSFWBMJEBUJPO FJUIFS EFDJTJPOXJMMOPUPOMZEFQFOEPOUIFCVEHFUB
UPVTFJOIPVTFTUBGGPSDPOUSBDUTUBGG TFF'JHVSF
GJSNJTDPNNJUUJOHUPUIJTFYFSDJTF CVUBMTPPOUIF
MFWFMPGFYQFSUJTFJOUIFGJSN'PSFYBNQMF TPNF
In-House Staff CJPUFDIOPMPHZTUBSUVQTNBZMBDLUIFFOHJOFFSJOHPS
5XPBQQSPBDIFTDBOCFUBLFOJOVTJOHJOIPVTFTUBGG SFHVMBUPSZFYQFSJFODFOFDFTTBSZUPQFSGPSNWBMJEBUJPO
POBSFWBMJEBUJPOQSPKFDU5IFGJSTUJTUPVTFUIFSFHVMBS BDUJWJUJFTGPSDPNNFSDJBMQSPEVDUJPO JOXIJDIDBTFB
WBMJEBUJPOTUBGGUPQFSGPSNUIJTUBTLCFTJEFUIFJS DPOTVMUBOUDPOUSBDUPSNBZCFUIFPOMZPQUJPO
EBJMZWBMJEBUJPOEVUJFT8IJMTUUIFEJSFDUDPTUPGTVDI 0ODFUIFGJSTUSFWBMJEBUJPOJTDPNQMFUFEBOEBMM
BOBQQSPBDIJTMPXDPNQBSFEUPPUIFST FYQFSJFODF UIFQSPEVDURVBMJUZJNQBDUJOHTZTUFNTIBWFCFFO
JOEJDBUFTUIBUJUJTIBSEUPFTUBCMJTIGPDVTBOEUP CSPVHIUJOUPDPNQMJBODF JUJTJNQFSBUJWFUPLFFQ
NBJOUBJODPNNJUNFOU5IFTFDPOEBQQSPBDIJTUP UIFNJOUIFWBMJEBUFETUBUF5IJTXJMMOFDFTTJUBUF
GPSNBOBEIPDDPNNJUUFFUPIBOEMFUIFSFWBMJEBUJPO SFHVMBS JOBOZPGUIFTFOTFTNFOUJPOFEFBSMJFS

QSPKFDUVOUJMJUTDPNQMFUJPO5IJTEPFTQSPWJEFCFUUFS SFWBMJEBUJPOPSSFFWBMVBUJPOPGUIPTFTZTUFNT5IJT
GPDVTBOENPSFDPNNJUNFOUCVUTUJMMUBLFTBMPOHUJNF DBOCFBDIJFWFECZNBLJOHUIFSFFWBMVBUJPOQBSU
UPBDIJFWFUIFUBSHFU PGUIFSFHVMBSTUBGGUBTLT PSCZFNQMPZJOHTUBGG
EFEJDBUFETPMFMZUPNBOBHJOHUIFSFWBMJEBUJPOBOE
Contract Staff the change control processes.
"HBJOUIFSFBSFUXPQPTTJCMFBQQSPBDIFT5IFGJSTU *OHFOFSBMUIFSFBSFTFWFSBMNPEFMTGPSVUJMJ[JOH
JTUPFNQMPZDPOUSBDUTUBGGPOBOJOEJWJEVBMCBTJT QFSTPOOFMGPSQFSGPSNJOHUIFWBMJEBUJPOGVODUJPO
BOENBOBHFUIFNCZBDPNQFUFOUQFSTPOGSPN JOBOPSHBOJ[BUJPOEFQFOEJOHPOXIFSFWBMJEBUJPO
UIFJOIPVTFTUBGG5IJTQFSTPOXJMMIBWFUPIBWF SFTQPOTJCJMJUJFTMJF'PSFYBNQMF XIFOWBMJEBUJPO
CPUIBEFRVBUFUFDIOJDBMBCJMJUJFTBOEFYDFMMFOU JTQBSUPGQSPDFTTEFWFMPQNFOU TUBGGNBZCFBMJHOFE
NBOBHFSJBMTLJMMT BTUIFZXJMMIBWFUPNBOBHFBO XJUIQSPEVDUTPSQSPEVDUDMBTTFT8IFOQBSUPG
JOIPNPHFOFPVTUFBNPGJOEJWJEVBMTPGEJGGFSFOU FOHJOFFSJOH UIFZBSFBMJHOFEFJUIFSXJUIGBDJMJUJFT
QSPGFTTJPOBM JOEVTUSJBM FEVDBUJPOBM BOEQPTTJCMZ TVDIBTDFSUBJOTFDUJPOTPGBTJUF PSXJUIMJOFTBOE
DVMUVSBMCBDLHSPVOET TVDIBTUIFDBTFJOUIF FRVJQNFOU"MUFSOBUJWFMZ WBMJEBUJPODBOCFQBSU
GSFFNPWFNFOUPGMBCPSXJUIJOUIF&6
5IJTXJMM PGUIFRVBMJUZBTTVSBODF 2"
GVODUJPO JOXIJDI
PCWJPVTMZCFDPTUMJFSUIBOIBWJOHUIFJOIPVTFTUBGG DBTFWBMJEBUJPOQFSTPOOFMUFOEUPXPSLPOEJGGFSFOU
QFSGPSNUIFKPC CVUXJMMFOUBJMIJHIFSDPNNJUNFOU QSPKFDUTJOWPMWJOHEJGGFSFOUQSPEVDUT GBDJMJUJFT
BOEBEFGJOFEUJNFGSBNF BOEFRVJQNFOU$MFBOJOH NJDSPCJPMPHJDBM BOE
The second option is to contract out the BOBMZUJDBMNFUIPETWBMJEBUJPOBSFOPSNBMMZXJUIJO
SFWBMJEBUJPOKPCXIFSFCZ BWBMJEBUJPODPOTVMUBOU UIFSFNJUPGUIFRVBMJUZDPOUSPMMBCPSBUPSJFT
DPOUSBDUPSXJMMIBOEMFUIFKPCBMUPHFUIFSXJUI
POMZBMJBJTPOQFSTPOGSPNUIFPXOFSTJEF*O CONCLUSION
UFSNTPGEJSFDUDPTUT UIJTPQUJPOJTOPSNBMMZUIF 5IJTEJTDVTTJPOEFTDSJCFEBTUSBUFHZGPSQMBOOJOH
DPTUMJFTUFTQFDJBMMZJGBTTFTTNFOUPGUIFWBMJEBUJPO BOEFYFDVUJOHSFWBMJEBUJPOPGFYJTUJOHTZTUFNT XJUI
TUBUVTJTJOUIFSFNJUPGUIFDPOUSBDUPS*OBEEJUJPO QBSUJDVMBSFNQIBTJTPOUIFGJSTUSFWBMJEBUJPO5IJT
UIFSFBSFJOEJSFDUDPTUTJOWPMWFEJOBVEJUJOHBOE FNQIBTJTJTCBTFEPOUIFBTTVNQUJPOUIBUUIFTZTUFN
BQQSPWJOHTVDIDPOUSBDUPSTBTQBSUPGBGJSNT IBTFJUIFSESJGUFEUPPNVDIGSPNJUTPSJHJOBMMZ
TVQQMJFSBVEJUQSPDFEVSFT0OUIFPUIFSIBOE UIF WBMJEBUFETUBUVTPSUIBUUIJTTUBUVTDBOOPUCFSFMJBCMZ
PXOFSXJMMHFUIJHIFSDPNNJUNFOUXJUISFTQFDU WFSJGJFE*GBMMUIFOFDFTTBSZEPDVNFOUBUJPOBOE
UPUJNF BOEDBOTUJQVMBUFQFOBMUJFTJOUIFDPOUSBDU JOGPSNBUJPOBSFBWBJMBCMFBOEWBMJEGSPNUIFPSJHJOBM
JGNJMFTUPOFTPSUJNFPCMJHBUJPOTBSFOPUNFU WBMJEBUJPO BOFWBMVBUJPOPGUIFDVSSFOUWBMJEBUJPO
'VSUIFSNPSF UIFDPOUSBDUPSCZWJSUVFPGXPSLJOHPO TUBUVTTIPVMECFTVGGJDJFOU$POTFRVFOUMZ GPSB
EJGGFSFOUQSPKFDUTGPSEJGGFSFOUDVTUPNFSTCSJOHTJO XFMMNBJOUBJOFEBOEDPOUSPMMFEQSPDFTT UIFDPTUPG
TJHOJGJDBOUFYQFSJFODFUPUIFGJSN5IFEPXOTJEF SFWBMJEBUJPOXPVMECFNJOJNBM

Figure 4: Business issues related to revalidation. 8)0 Supplementary Guidelines on Good Manufacturing
Practices (GMP): Validation 0DUPCFS BWBJMBCMFBU
Revalidation costs
www.who.int.
'%" Guidance for Industry Part 11, Electronic Records;
Electronic SignaturesScope and Application "VHVTU
BWBJMBCMFBUXXXGEBHPW
Assessment costs Performance costs
'%" Guidance for Industry Process Validation: General
Principles and Practices, Draft Guidance BWBJMBCMF
Revalidation stafng
BUXXXGEBHPW
*41& Baseline Pharmaceutical Engineering Guide: Volume
5 Commissioning & Qualification
*$) Q7 Good Manufacturing Practice Guide for Active
In-house Contract
Pharmaceutical Ingredients BWBJMBCMFBUXXXJDI
PSH"MTPJODPSQPSBUFEJOUP7PMVNFPGUIF&6(.1BT
QBSU**JO BOEBTBHVJEBODFGPSJOEVTUSZCZ'%"
JO
Regular staff Ad-hoc committee Individuals Firm
'%" Pharmaceutical CGMPs for the 21st Century: A Risk-
based Approach i"TDJFODFBOESJTLCBTFEBQQSPBDIUP
Time Cost Committment QSPEVDURVBMJUZSFHVMBUJPOJODPSQPSBUJOHBOJOUFHSBUFE
In-house Long/Open Low low
RVBMJUZTZTUFNTBQQSPBDI wBWBJMBCMFBUXXXGEB
HPW
Contract Denite High High
&(SBOUBOE3-FBWFOXPSUI Statistical Quality Control
UIFE .D(SBX)JMM
$IBOHFNBOBHFNFOUJTBLFZDPNQPOFOUUP &MB JOF .B S U JO +VMJB O .PS S JT B OE 4UF WFO - B OF
UIFSFWBMJEBUJPOTUSBUFHZ$IBOHFNBOBHFNFOU i.POJUPSJOH1SPDFTT.BOVGBDUVSJOH1FSGPSNBODF wIEEE
EFUFSNJOFTUIFOFFEGPSSFWBMJEBUJPOBOE Control Systems Magazine 0DUPCFSQQoJVT
TVCTFRVFOUMZHVBSBOUFFTNBJOUBJOJOHUIFBDRVJSFE
WBMJEBUFETUBUVT"OPUIFSLFZDPNQPOFOUJTUIF ARTICLE ACRONYM LISTING
TUSVDUVSFEBQQSPBDIUPWBMJEBUJPONBJOUFOBODFBT API Active Pharmaceutical Ingredient
UJNFCBTFEBOEFWFOUCBTFEXJUITVCDBUFHPSJFTPG EHS Environmental, Health, and Safety
FBDI8FCFMJFWFUIJTBQQSPBDIUPCFBQSBHNBUJD EU European Union
BOESBUJPOBMPOF*UTNBJOFTTFODFJTUPCBTFUIF FAT Factory Acceptance Testing
key decisions on the potential risk associated with FDA US Food and Drug Administration
UIFEJGGFSFOUBDUJWJUJFT NBJOMZSJTLUPUIFQBUJFOU FMEA Failure Mode Effect Analysis
CVUBMTPJODPSQPSBUJOHSJTLUPUIFCVTJOFTTXIFO GMP Good Manufacturing Practice
appropriate. It is not suggested that the features ICH International Conference on Harmonisation
BOEGVODUJPOTUPCFWBMJEBUFEJOBTZTUFNBSFUP IPR Intellectual Property Right
CFEFUFSNJOFETPMFMZCZCVTJOFTTJTTVFT3BUIFS ISO International Organization for
UIFCVTJOFTTGBDUPSTPOMZEFUFSNJOFUIFPSEFSJO Standardization
XIJDIUIFTZTUFNTBSFUPCFSFWBMJEBUFEUIPTF ISPE International Society for Pharmaceutical
BXBJUJOHSFWBMJEBUJPOIBWFPUIFSBMUFSOBUJWFT Engineering
6OEFSOPDJSDVNTUBODFTXPVMEUIFQSPEVDUPGB MSPC Multivariate Statistical Process Control
NBOVGBDUVSJOHQSPDFTTUIBUJTOPUWBMJEBUFEUP PAT Process Analytical Technology
SFHVMBUPSZFYQFDUBUJPOTCFSFMFBTFE PIC/S Pharmaceutical Inspection Convention/
Pharmaceutical Inspection Co-Operation
REFERENCES QA Quality Assurance
*$) Q9 Quality Risk Management BWBJMBCMFBU R&D Research and Development
www.ich.org. SAT Site Acceptance Testing
&VSPQFBO$PNNJTTJPO The Rules Governing Medicinal SPC Statistical Process Control
Products in the European Union 7PMVNF (PPE URS User Requirement Specification
.BOVGBDUVSJOH1SBDUJDFT &EJUJPOVQEBUFEJO WHO World Health Organization
BWBJMBCMFBUFDFVSPQBFVFOUFSQSJTFQIBSNBDFVUJDBMT
eudralex.
0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of Validation Technology

John McConnell, Brian K. Nunnally, and Bernard McGarvey
Dos and Donts of Control

ChartingPart I
Analysis and Control of Variation is dedicated B

"TBOBOBMZTJTUPPMUPVOEFSTUBOEUIFIJTUPSJDBM
to revealing weaknesses in existing approaches to CFIBWJPSPGUIFQSPDFTT MPPLGPSQFSJPETPGTUBCMF
understanding, reducing, and controlling variation BOEVOTUBCMFCFIBWJPVS
and to recommend alternatives that are not only C
#BTFEPOUIFSFTVMUTPG B
EFGJOJOHXIBUTUBCMF
based on sound science, but that demonstrably work. WBSJBUJPOMPPLTMJLFBOEUIFOVTJOHUIFDPOUSPM
The objective of the column is to combine sound chart as a control tool to assess if the process is
science with proven practical advice. DVSSFOUMZTUBCMFPSOPUBOEXIBUBDUJPOT JGBOZ
Reader comments, questions, and suggestions will TIPVMECFUBLFOCBTFEPOUIJTBTTFTTNFOU
help us fulfil our objective for this column. Case
studies illustrating the successful reduction or control 5IVT B
JTUIFGVOEBNFOUBMBOBMZTJTUIBUJTEPOF
of variation submitted by readers are most welcome. XJUIBDPOUSPMDIBSU)PXFWFS MJLFBOZUIJOHFMTF
Please send your comments and suggestions to column there are right ways and wrong ways to set up and
coordinator John McConnell at john@wysowl.com. use control charts. It is the purpose of this article
au or journal coordinating editor Susan Haigney at to explore the right and wrongs of control charting.
shaigney@advanstar.com. 5IJTUPQJDJTQSFTFOUFEJOUXPQBSUT*OUIJTGJSTUQBSU
FYBNQMFTBSFVTFEUPJMMVTUSBUFXIBUDBOIBQQFOJG
KEY POINTS DISCUSSED DPOUSPMDIBSUTBSFOPUVTFEDPSSFDUMZ5IFTFFYBNQMFT
The following key points are discussed in this article: BSFCBTFEPOSFBMTJUVBUJPOTXJUIXIJDIUIFBVUIPST
tThere is a right way and a wrong way to IBWFCFFOJOWPMWFE*OUIFTFDPOEQBSU XIJDIXJMM
control chart CFBGVUVSFBSUJDMFJOUIJTTFSJFTPOUIFi"OBMZTJTBOE
t"MXBZTTUBSUUIFBOBMZTJTPGEBUBCZMPPLJOHBU $POUSPMPG7BSJBUJPO wBNPSFDPNQSFIFOTJWFTFUPG
UIFUSFOEPWFSUJNF DPOUSPMDIBSUJOHHVJEFMJOFTXJMMCFEJTDVTTFE
t.BLFTVSFUIBUUIFNFUSJDCFJOHBOBMZ[FEJT
UIFCFTUXBZBWBJMBCMFUPSFQSFTFOUUIFEJSFDU EXAMPLE 1
WBSJBUJPOPGUIFTZTUFN "QSPDFTTJNQSPWFNFOUUFBNXBTDIBSHFEXJUI
t%BUBBSFHPME.BLFTVSFZPVVTFBMMUIF SFEVDJOHUIFOVNCFSPGQSPEVDUJPOJTTVFTJOBQSPDFTT
BWBJMBCMFEBUBCFGPSFSFBDIJOHBDPODMVTJPO 5IJTQSPDFTTDPOTJTUFEPGGJWFTUFQT"TFYQFDUFE UIF
GJSTUUIJOHUIFUFBNEJEXBTUPHBUIFSEBUBPOUIF
INTRODUCTION historical occurrence of these production issues. They
"QSFWJPVTBSUJDMF
JOUIJTTFSJFTEJTDVTTFTUIFOFFEUP UIFOQMPUUFEUIFEBUBBTB1BSFUPDIBSUBTTIPXOJO
VTFDPOUSPMDIBSUJOHBTBOJOUFHSBMQBSUPGBOZRVBMJUZ 'JHVSF
JNQSPWFNFOUTZTUFN$POUSPMDIBSUTDBOCFVTFEJO 5IF1BSFUPDIBSUJOEJDBUFEUIFHSFBUFTUOVNCFS
UIFGPMMPXJOHUXPNPEFT
PGQSPEVDUJPOJTTVFTPDDVSSFEXJUITUFQPGUIF
[
ABOUT THE AUTHORS
For more Author John McConnell is owner and director at Wysowl Pty Ltd. in Queensland, Australia. He may be
information, reached at john@wysowl.com.au. Brian K. Nunnally, Ph.D., is in charge of process validation at
go to
Pfizer in Sanford, NC. He may be reached at nunnalb@wyeth.com. Bernard McGarvey, Ph.D., is
Process Modeling Group Leader, Process Engineering Center at Eli Lilly and Company in Indianapolis,
gxpandjvt.com/bios
IN. He may be reached at mcgarvey_bernard@lilly.com.

QSPEVDUJPOQSPDFTT#BTFEPOUIJTDIBSU UIFUFBN
Figure 1: Pareto of production issues.
EFDJEFEUPTUBSUMPPLJOHBUTUFQJONPSFEFUBJM
GPMMPXFECZTUFQBOETPPO'PSHFUUJOHUIFRVFTUJPO
BTUPXIFUIFS'JHVSFSFQSFTFOUTBHPPE1BSFUP
TJODFPGUIFJTTVFTBSFOPUDBVTFECZPG
UIFTUFQT
UIFUFBNIBEOFHMFDUFEUPGPMMPXXIBUJT
Cum percent
QSPCBCMZUIFNPTUJNQPSUBOUSVMFPGEBUBBOBMZTJT
Count
"MXBZTTUBSUBOZEBUBBOBMZTJTCZQMPUUJOHUIFEBUB
PWFSUJNFUPTFFJGUIFEBUBJOEJDBUFBTUBCMFTZTUFN
8IFOUIFEBUBXFSFVTFEUPDSFBUFBUSFOETIPXJOH
UIFOVNCFSPGJTTVFTQFSNPOUI 'JHVSFXBTUIF
result.
'JHVSFJOEJDBUFTBHSFBUJODSFBTFJOQSPEVDUJPO
JTTVFTBUBTQFDJGJDUJNFQFSJPEPWFSUIFDPVSTFPGUJNF
$MFBSMZTPNFUIJOHTJHOJGJDBOUIBQQFOFECFUXFFO
NPOUITBOEUIBUSFTVMUFEJOBOJODSFBTFJOUIF
OVNCFSPGJTTVFT5IJTNFBOTUIBUUIF1BSFUPDIBSU
Step#
TIPXOJO'JHVSFJTNFBOJOHMFTT/PUPOMZEPFTJU
JODMVEFEBUBGSPNUXPWFSZEJGGFSFOUQSPDFTTFT CVUJU
DPNQMFUFMZIJEFTUIFGBDUUIBUBNBKPSQSPDFTTDIBOHF
IBEPDDVSSFE3BUIFSUIBOGPDVTTJOHPOUIF1BSFUP Figure 2: Trend of production issues over time.
UIFUFBNTIPVMEMPPLGPSUIFSPPUDBVTFPGXIZUIF
DIBOHFPDDVSSFE0OFXBZUPTUBSUUIJTJOWFTUJHBUJPOJT
Issues per month
UPDPNQBSFUIF1BSFUPTGPSFBDIPGUIFSFHJPOTJO'JHVSF
SFGFSSJOHUPUIFGJSTUBTSFHJPO"BOEUIFTFDPOEBT
SFHJPO# BTTIPXOJO'JHVSF
/PUJDFUIBUDPNQBSJOHUIFDIBSUTJO'JHVSFTIPXT
UIBUUIFCJHHFTUDIBOHFCFUXFFOUIFUXPSFHJPOTXBT
BTJHOJGJDBOUJODSFBTFJOJTTVFTGPSTUFQ4UFQIBE
BHSFBUJODSFBTFJOSFHJPO#DPNQBSFEUPBWFSZMPX
MFWFMJOSFHJPO"*OWFTUJHBUJPOPGTUFQTIPVMECF Time (months)
UIFJOJUJBMGPDVTPGUIFJNQSPWFNFOUUFBN0ODFUIF
SPPUDBVTFGPSUIJTJODSFBTFIBTCFFOJEFOUJGJFEBOE
FMJNJOBUFE UIFUFBNTIPVMECFCBDLUPUIFTJUVBUJPO
Figure 3: Pareto of production issues by region.
GPSSFHJPO"5IFSFGPSF UIFJSOFYUTUFQXPVMECFUP
MPPLBUTUFQ5IVTCZVTJOHUIFUSFOEBOBMZTJTGJSTU
UIFUFBNUPPLBDPNQMFUFMZEJGGFSFOUBQQSPBDIUP
JNQSPWFNFOU POFXIJDIXPVMEZJFMESFTVMUTBMPU
GBTUFSUIBOUIFJSPSJHJOBMQMBOCBTFEPOBGMBXFE1BSFUP
Cum percent
analysis.
Count
5IFMFTTPOUPCFMFBSOFEIFSFJTUPBMXBZTTUBSUCZ
MPPLJOHBUUIFUSFOEPGEBUBPWFSUJNF
EXAMPLE 2
"OJNQPSUBOUNFBTVSFPGQFSGPSNBODFGPSBOZ
Step# Step#
QSPEVDUJPOTZTUFNJTUIFBCJMJUZPGUIFTZTUFNUPBEIFSF
to its production schedule. Typically the production
TZTUFNJTCSPLFOVQJOUPBOVNCFSPGTUFQT VTVBMMZ QSPHSBNJOBQSPEVDUJPOBSFB SFGFSSFEUPBT%$
UIF
BTUFQJTBOJOEJWJEVBMXPSLTUBUJPO
BOEUIFTDIFEVMF QSPDFTTJNQSPWFNFOUUFBNEFDJEFEUPJNQMFNFOUB
specifies when lots are supposed to start at each step. DPOUSPMDIBSUGPSUIJTNFUSJD5IJTXPVMEBMMPXUIF
"EIFSFODFUPTDIFEVMFJTNFBTVSFECZMPPLJOHBUUIF UFBNUPBTTFTTUIFTUBUFPGDPOUSPMGPSUIJTNFUSJDBOE
UPUBMOVNCFSPGMPUTUIBUBSFCFIJOETDIFEVMFBUBMM EFQFOEJOHPOUIFSFTVMUTPGUIJTBTTFTTNFOU UPVTFUIJT
UIFTUFQT SFGFSSFEUPBTOVNCFSPGQBTUEVFTUFQT
UPDSFBUFJNQSPWFNFOUBDUJPOQMBOT5IFQFSGPSNBODF
5IJTNFBOTUIBUUIFNFUSJDDBOCFCJHHFSUIBOUIF EBUBGPSUIFNFUSJDXFSFTVQQMJFEJOUIFGPSNPGBUSFOE
UPUBMOVNCFSPGTUFQTJOUIFQSPEVDUJPOTZTUFN/PX DIBSUPWFSBCPVUBTFWFONPOUIQFSJPE5IJTDIBSUJT
BTQBSUPGBOJNQMFNFOUBUJPOPGBOJNQSPWFNFOU TIPXOJO'JHVSF

$MFBSMZTPNFUIJOHXBTHPJOHPO CVUJUXBTOPUDMFBS
Figure 4: Trend of missed steps.
IPXUPJOUFSQSFUUIJTUSFOEBOEIPXCFTUUPDPOUSPM
DIBSUUIFEBUB*TUIJTSFBMMZUIFCFTUXBZUPDIBSUUIJT
Past due steps

NFUSJD 5IFSFBTPOGPSBTLJOHUIJTRVFTUJPOJTCBTFEPO
XIBUHPFTJOUPUIJTNFUSJD*UJTDMFBSUIBUUIFOVNCFS
PGQBTUEVFTUFQTJTJOGMVFODFECZUIFGPMMPXJOHUXP
DPNQPOFOUT
t*UJODSFBTFTBTOFXMPUTBSSJWFUIBUBSFOPU
QSPDFTTFEPOUJNF
tIt decreases as lots get processed at a step.
*GUIFWBMVFPGUIFNFUSJDBUUJNFUJTEFOPUFEBT.t Dates
then we can say
Figure 5: Auto-correlation plots.
.t.U + Nto0t
<&RVBUJPO> DC daily correlation chart
35
XIFSF/JTUIFOVNCFSPGOFXMBUFTUFQTCFJOH 30
BEEFEBOE0JTUIFOVNCFSPGMBUFTUFQTUIBUIBWF 25
CFFODPNQMFUFEBOETPBSFOPMPOHFSDPOTJEFSFEMBUF
Mt 20
*UDBOOPXCFTFFOUIBUUIFOVNCFSPGQBTUEVFTUFQT
JTMJLFBOJOWFOUPSZ5IFBDUVBMQFSGPSNBODFPGUIF 15
QSPDFTTJTSFBMMZTIPXOEJSFDUMZCZUIFWBSJBUJPOPG/ 10
BOE0 XIFSFBT.TIPXTUIFWBSJBUJPOJOBNVDINPSF
5
JOEJSFDUXBZ)PXFWFS JGXFSFXSJUF&RVBUJPOBT
0
0 5 10 15 20 25 30 35
.t.t.U = Nto0t Mt-1
<&RVBUJPO>
DC daily correlation chart
10
8FDBOTFFUIBUUIFEJGGFSFODFJOUIFNFUSJDWBMVFT
EJSFDUMZSFGMFDUTUIFWBSJBUJPOPG/BOE0BMUIPVHI 5
JUDBOOPUTIPXUIFWBSJBUJPOPG/BOE0JOEJWJEVBMMZ
5IJTOFXNFUSJD .t JTSFGFSSFEUPBTUIFEBJMZEFMUB 0
-15 -10 -5 10 15
NFUSJD0OFBEWBOUBHFPGUIJTNFUSJDJTUIBUJGJUUSVMZ
Mt
SFQSFTFOUTUIFQSPDFTTQFSGPSNBODFNPSFEJSFDUMZUIBO -5
UIFPSJHJOBMNFUSJD BOBVUPDPSSFMBUJPOQMPUTIPVME
TIPXMFTTDPSSFMBUJPOGPSUIF.NFUSJDUIBOUIF. -10
NFUSJD'JHVSFTIPXTUIFBVUPDPSSFMBUJPODIBSUTGPS
UIFPSJHJOBMNFUSJDBOEUIFOFXNFUSJD -15
Mt-1
*UJTDMFBSUIBUUIFEBJMZEFMUBNFUSJDIBTOPBVUP
DPSSFMBUJPO XIFSFBTUIFPSJHJOBMNFUSJDTIPXT
TJHOJGJDBOUBVUPDPSSFMBUJPO XIJDIJTXIBUXFXPVME There is also a hint that at the end of the chart the
FYQFDU0OFPGUIFBTTVNQUJPOTPGUIFCBTJDDPOUSPM WBSJBCJMJUZNBZCFSJTJOHBHBJO BTJOEJDBUFECZUIF
DIBSUTTVDIBTBOJOEJWJEVBMNPWJOHSBOHFDIBSUJT PVUPGDPOUSPMTJHOBMPOUIF.PWJOH3BOHFDIBSU5IJT
UIBUUIFEBUBCFJOHQMPUUFEBSFOPUTJHOJGJDBOUMZBVUP JTTPNFUIJOHUIFUFBNOFFETUPLFFQPOUIFJSSBEBS
DPSSFMBUFE5IFSFGPSF UIF.NFUSJDJTTVJUBCMFGPS screen.
TVDIBDIBSU XIFSFBTUIFPSJHJOBMNFUSJDJTOPU/PX 5IFMFTTPOMFBSOFEIFSFJTUPNBLFTVSFUIBUUIF
UIFEBJMZEFMUBNFUSJDDBOCFQMPUUFEVTJOHBTUBOEBSE NFUSJDCFJOHBOBMZ[FEJTUIFCFTUXBZBWBJMBCMFUP
JOEJWJEVBMTNPWJOHSBOHFDIBSU BTTIPXOJO'JHVSF SFQSFTFOUUIFEJSFDUWBSJBUJPOPGUIFTZTUFN
'JHVSFTIPXTBWFSZDMFBSQJDUVSFPGXIBUIBTCFFO
HPJOHPOPWFSUIFTFWFONPOUIQFSJPE5IFBWFSBHF EXAMPLE 3
IBTOPUDIBOHFEPWFSUIJTQFSJPECVUUIFWBSJBCJMJUZ "QSPDFTTJNQSPWFNFOUUFBNXBTHJWFOUIFPCKFDUJWF
PGUIFTZTUFNIBTDPNFEPXOWFSZTJHOJGJDBOUMZ PGJNQSPWJOHUIFQSPEVDUJPOSBUFJOBNBOVGBDUVSJOH
5IFQSPDFTTJNQSPWFNFOUUFBNTIPVMEGPDVTUIFJS TZTUFN*OPSEFSUPEPUIJT UIFZHBUIFSFEUIFEBJMZ
attention on understanding what has caused this and production rate data and plotted the data as an
XIFUIFSUIJTSFEVDUJPOJOWBSJBCJMJUZDBOCFTVTUBJOFE JOEJWJEVBMNPWJOHSBOHFDIBSUBTTIPXOJO'JHVSF

Figure 6: Control charts for the daily delta Figure 7: Control chart for production rate.
metric.
X chart for DC missed step delta
Daily production
Individual values
Date
Date
mR chart for DC missed step delta

Moving range values
Figure 8: Control chart for production rate

by shift.
X chart for shift B pen production per shift
Individual values
Date
$MFBSMZUIFSFJTBMPUPGWBSJBCJMJUZJOUIFQSPEVDUJPO
TZTUFN4PUIFUFBNIFMEBCSBJOTUPSNJOHTFTTJPOUP
Date
USZUPDPNFVQXJUIBOJOJUJBMMJTUPGQSPCMFNBSFBT
UIBUNJHIUCFDBVTJOHUIJTWBSJBCJMJUZ0OFUIJOHUIBU X chart for shift C pen production per shift
Individual values
DBVHIUUIFJSBUUFOUJPOXBTUIFJTTVFPGWBSJBUJPOEVF
UPUIFQFPQMFTZTUFN#FDBVTFUIFDPNQBOZIBEWFSZ
HPPEQSPDFEVSFTBOETPPO UIFJNQSPWFNFOUUFBN
GFMUUIBUWBSJBUJPOEVFUPUIFQFPQMFTZTUFNXBTWFSZ
MPXBOEJOGBDUXBTBUUIFCPUUPNPGUIFJSMJTU'VSUIFS
EJTDVTTJPOSFWFBMFEUIBUUIFQSPEVDUJPOSBOBOE
XBTDPWFSFECZTIJGUT BOEUIFZIBEUIFQSPEVDUJPO Date
EBUBCZTIJGUGPSFBDITIJGU4PJUXBTTVHHFTUFEUIBUUIF
EBUBCFDIBSUFECZTIJGUUPTFFJGUIFSFXBTBOZTIJGUUP GPSJNQSPWFNFOUJOIPXTVDIDIBSUTBSFVTFE5IJT
TIJGUWBSJBUJPO'JHVSFTIPXTUIFSFTVMUJOHDIBSUTGPS BSUJDMFIBTTIPXOKVTUUISFFTJNQMFFYBNQMFTXIFSF
KVTUUXPPGUIFTIJGUT UIFJOJUJBMBQQSPBDICFJOHVTFEXBTMFTTUIBOPQUJNBM
$MFBSMZUIFSFJTWFSZEJGGFSFOUCFIBWJPVSCFUXFFO BOEIPXBOJNQSPWFEDPOUSPMDIBSUJOHBQQSPBDIXBT
UIFTFUXPTIJGUT BOEUIFPUIFSUISFFBTXFMM
"MTP UIF VTFEUPSFWFBMBCFUUFSQJDUVSFPGXIBUJTSFBMMZHPJOH
EBUBTIPXTJHOJGJDBOUDIBOHFTXJUIJOFBDITIJGU CVU PO8JUIPVUBUSVFVOEFSTUBOEJOHPGXIBUJTHPJOHPO
UIFTFDIBOHFTXFSFIBQQFOJOHBUEJGGFSFOUUJNFTGPS BOZJNQSPWFNFOUQSPHSBNXJMMCFGMBXFE"GVUVSF
FBDITIJGU5IFJNQSPWFNFOUUFBNXBTUBLFOBCBDL BSUJDMFJOUIJTTFSJFTPOUIFi"OBMZTJTBOE$POUSPMPG
when they saw these charts and without hesitation 7BSJBUJPOwXJMMMPPLBUTPNFPUIFSXBZTUIBUDPOUSPM
UIFZNPWFEWBSJBUJPOEVFUPQFPQMFGSPNUIFCPUUPN DIBSUTDBOCFNJTVTFEBOEXBZTUPBWPJETVDINJTVTF
of the list to the top of the list.
5IFMFTTPOMFBSOFEIFSFJTEBUBBSFHPME.BLF
TVSFZPVVTFBMMUIFBWBJMBCMFEBUBCFGPSFSFBDIJOHB REFERENCES
conclusion. .D$POOFMM +PIO /VOOBMMZ #SJBO BOE .D(BSWFZ
#FSOBSE i7BSJBUJPO1BTU 1SFTFOU BOE'VUVSF wJournal of
CONCLUSIONS Validation Technology 7PMVNF/VNCFS 4QSJOH
$POUSPMDIBSUJOHJTBCBTJDUPPMJOUIFKPVSOFZPG .D$POOFMM +PIO Analysis and Control of Variation 'PVSUI
RVBMJUZJNQSPWFNFOU)PXFWFS FWFOUIPVHIXF &EJUJPO %FMBXBSF#PPLT JVT
BSFJOUIFTUDFOUVSZBOEDPOUSPMDIBSUTXFSF
JOWFOUFECBDLJOUIFT UIFSFJTTUJMMNVDISPPN

John McConnell, Brian Nunnally, and Bernard McGarvey
The Dos and Donts of Control

ChartingPart II
Analysis and Control of Variation is dedicated INTRODUCTION

to revealing weaknesses in existing approaches "QSFWJPVTBSUJDMF
JOUIJTTFSJFTEJTDVTTFTUIF
to understanding, reducing, and controlling GPMMPXJOHEPTBOEEPOUTBTTPDJBUFEXJUIUIFVTFPG
variation and to recommend alternatives that control charting:
are not only based on sound science, but also t%PBMXBZTTUBSUCZMPPLJOHBUUIFUSFOEPGEBUB
that demonstrably work. Case studies will be PWFSUJNF
used to illustrate both problems and successful t%POPUVTFBNFUSJDUIBUJTOPUUIFCFTUXBZ
methodologies. The objective of the column is BWBJMBCMFUPSFQSFTFOUUIFEJSFDUWBSJBUJPOPGUIF
to combine sound science with proven practical TZTUFN
advice. t%BUBBSFHPMEEPNBLFTVSFZPVVTFBMMUIF
Reader comments, questions, and suggestions BWBJMBCMFEBUBCFGPSFSFBDIJOHBDPODMVTJPO
will help us fulfil our objective for this column. t& YBNQMFDBTFTUVEJFTJMMVTUSBUJOHUIFBCPWFBSF
Case studies illustrating the successful reduction QSPWJEFE
or control of variation submitted by readers are
most welcome. Please send your comments and 1BSUPOFPGUIJTBSUJDMFJMMVTUSBUFTUISFFFYBNQMFT
suggestions to column coordinator John McConnell BEESFTTJOHUIFEPTBOEEPOUTPGDPOUSPMDIBSUJOH
at john@wysowl.com.au or journal coordinating 1BSUUXPDPOUJOVFTUPJMMVTUSBUFUIFDPODFQUT
editor Susan Haigney at shaigney@advanstar.com. QSFTFOUFEJOQBSUPOFJOUIFGPMMPXJOHFYBNQMFT

KEY POINTS DISCUSSED EXAMPLE FOUR

The following key points are discussed: 4PNFUJNFTXFNPOJUPSBNFBTVSFBUBMPXFSGSFRVFODZ
tThere is a right way and a wrong way to control UIBOJTQPTTJCMF5IJTSFEVDFTUIFVTFGVMOFTTPGUIF
chart. NFBTVSF5IFNPUJWBUJPOGPSUIJTJTVTVBMMZUIBUXF
t%POPUVTFBHHSFHBUFEEBUBXIFODPOTUSVDUJOHB BMSFBEZIBWFXFFLMZPSNPOUIMZNFFUJOHTUIBUSFWJFX
control chart. NFBTVSFTBOEXFTJNQMZPSHBOJ[FUIFEBUBBSPVOE
t%POPUDPOUSPMDIBSUBNFBTVSFTJNQMZCFDBVTF UIJTGSFRVFODZ4PXFDPOUSPMDIBSUEBUBUIBUBSF
JUJTFBTZ$POUSPMDIBSUUIFNFBTVSFCFDBVTFJU BHHSFHBUFEGSPNBNPSFHSBOVMBSTFUPGEBUB"MTP
HJWFTJOTJHIUJOUPUIFQSPDFTT we reason that the people who are looking at the data
t%PNBLFTVSFUIBUUIFDPOUSPMDIBSUMJNJUTBSF XJMMOPUCFJOUFSFTUFEJOUIFEBUBPOBNPSFGSFRVFOU
calculated correctly. CBTJT)PXFWFS XIFOXFEPUIJT XFNBZMPTF
[
ABOUT THE AUTHORS
For more Author John McConnell is owner and director at Wysowl Pty Ltd. in Queensland, Australia. He may be
information, reached at john@wysowl.com.au. Brian K. Nunnally, Ph.D., is in charge of process validation at
go to
Pfizer in Sanford, NC. He may be reached at nunnalb@wyeth.com. Bernard McGarvey, Ph.D., is
Process Modeling Group Leader, Process Engineering Center at Eli Lilly and Company in Indianapolis,
gxpandjvt.com/bios
IN. He may be reached at mcgarvey_bernard@lilly.com.

Figure 1: Conversion time monthly average.
X Chart for Conversion Time Monthly Average

80
Individual Values
70
60
50
40
30
20
10
0
Jul-01
sep-01
Jan-02
Jun-01
Oct-01
Feb-01
Feb-02
Apr-01
Dec-01
Nov-01
Mar-01
Aug-01
May-01
Date
mR Chart for Conversion Time Monthly Average

35
Moving Range Values
30
25
20
15
10
5
0
Jul-01
sep-01
Jan-02
Jun-01
Oct-01
Feb-01
Feb-02
Apr-01
Dec-01
Nov-01
Mar-01
Aug-01
May-01
Date
TJHOJGJDBOUJOGPSNBUJPOBCPVUUIFDVSSFOUBOEQBTU CFDBVTFXFPOMZIBWFEBUBQPJOUT"QQSPYJNBUFMZ
CFIBWJPSPGUIFNFBTVSFCFJOHFWBMVBUFE5IJTDBO UISFFUPGPVSMPUTPGUIFQSPEVDUBSFNBEFQFSNPOUI
MFBEUPJODPSSFDUDPODMVTJPOTBCPVUXIBUXFTIPVME TPJGXFXFSFUPMPPLBUUIFEBUBPOBQFSMPUCBTJT XF
CFEPJOHXJUISFTQFDUUPUIJTNFBTVSF*OUIFQBTUXF XPVMEJNNFEJBUFMZHFUNVDINPSFEBUB5IFMPUCZMPU
NBZIBWFCFFOBCMFUPSBUJPOBMJ[FUIJTBQQSPBDIXIFO DPOUSPMDIBSUJTTIPXOJO'JHVSF
UIFDPOTUSVDUJPOPGDPOUSPMDIBSUTXBTNPSFMBCPS 8FOPXIBWFEBUBQPJOUTJOTUFBEPG5IF
JOUFOTJWF8JUINPEFSODPNQVUFSTZTUFNT IPXFWFS CFIBWJPSPGUIFDPOWFSTJPOUJNFJTOPXNVDIDMFBSFS
this is no longer the case. 8FIBWFIBECBTJDBMMZUISFFSFHJPOTPGCFIBWJPS
"TBOFYBNQMF XFMPPLBUBTFUPGEBUBSFQSFTFOUJOH PWFSUIFMBTUZFBS5IFTFBSFBTGPMMPXT
UIFDPOWFSTJPOUJNF DZDMFUJNF
GPSBMPUPGBDFSUBJO t3FHJPOGSPNMPUUPMPU5IJTIBTUXP
QSPEVDU5IFEBUBBSFBWFSBHFEPWFSBNPOUIMZUJNF TQFDJBMDBVTFQPJOUTBUMPUBOEMPU5IF
QFSJPEBOEBSFSFWJFXFEJOBNPOUIMZNFFUJOH5IF DPNNPODBVTFBWFSBHFBOETJHNBGPSUIJTSFHJPO
DPOUSPMDIBSUGPSUIFEBUBPWFSBNPOUIUJNFQFSJPE BSFBOEEBZT SFTQFDUJWFMZ
JTTIPXOJO'JHVSF t3FHJPOGSPNMPUUPMPU5IFSFBSFOP
'JHVSFTVHHFTUTUIBUUIFSFNJHIUCFBOVQXBSE TQFDJBMDBVTFTJHOBMT5IFDPNNPODBVTFBWFSBHF
USFOE CVUUIFEBUBBSFJODPODMVTJWF5IFEBUBEPOPU BOETJHNBGPSUIJTSFHJPOBSFBOEEBZT
CSFBLBOZTUBUJTUJDBMUSFOESVMF XIJDIJTOPUTVSQSJTJOH SFTQFDUJWFMZ

Figure 2: Product conversion time on a lot basis.
X Chart for Conversion Time by Lot

90
Individual Values
80
70
60
50
40
30
20
10
0
Lot 1
Lot 3
Lot 5
Lot 7
Lot 9
Lot 11
Lot 13
Lot 15
Lot 17
Lot 19
Lot 21
Lot 23
Lot 25
Lot 27
Lot 29
Lot 31
Lot 33
Lot 35
Lot 37
Lot 39
Date
mR Chart for Conversion Time by Lot

Moving Range Values
40
35
30
25
20
15
10
5
0
Lot 1
Lot 3
Lot 5
Lot 7
Lot 9
Lot 11
Lot 13
Lot 15
Lot 17
Lot 19
Lot 21
Lot 23
Lot 25
Lot 27
Lot 29
Lot 31
Lot 33
Lot 35
Lot 37
Date Lot 39
t3FHJPOGSPNMPUUPMPU*UBQQFBSTUIBU EXAMPLE FIVE

the last three to four lots of these data are trending 4PNFUJNFTXFDPOUSPMDIBSUNFBTVSFTUIBUEPOPU
sharply upward. SFBMMZNFBTVSFUIFJOUFOEFEQSPDFTTQFSGPSNBODF5IF
rational for this is typically a lack of understanding of
8FDBOTFFGSPN'JHVSFUIBUUIFUSFOEXF UIFUSVFQVSQPTFPGUIFQSPDFTTPSBSFBMJ[BUJPOUIBU
TVTQFDUFEJO'JHVSFMPPLTNPSFMJLFBTUFQDIBOHFBU BNFBTVSFUIBUUSVMZSFQSFTFOUTXIBUXFXBOUXJMMCF
MPU5IFSFGPSF XFTIPVMECFBTLJOHXIBUDIBOHFE EJGGJDVMUUPNFBTVSF8FUIFOVTFBTJNQMFSNFBTVSF
JOUIFQSPDFTTCFUXFFOMPUTBOE4JNJMBSMZ XF 4PNFFYBNQMFTEFNPOTUSBUJOHUIJTBSFBTGPMMPXT
TIPVMECFBTLJOHXIBUDIBOHFECFUXFFOMPUTBOE tPercent completed training. The only
BOEXIBUJTDBVTJOHUIFUSFOEVQXBSEGSPNMPUTBOE NFBTVSFXFIBWFBSPVOEQFPQMFQFSGPSNBODF
/POFPGUIFTFRVFTUJPOTXPVMECFBTLFEJGXFPOMZ JT*51*51JTUIFQFSDFOUPGUIFJSUSBJOJOH
MPPLBU'JHVSF QMBOUIBUJOEJWJEVBMT *51
IBWFDPNQMFUFE
5IVTXFBSFNJTTJOHBMBSHFBNPVOUPGQSPDFTT 5IJTNFBTVSFPOMZSFGMFDUTXIFUIFSBQFSTPO
CFIBWJPSCZPOMZMPPLJOHBUNPOUIMZBWFSBHFEBUB IBTPCUBJOFEBMMUIFJSSFRVJSFEUSBJOJOHDPVSTFT
5IFMFTTPOMFBSOFEGSPNFYBNQMFGPVSJTUIBU )PXFWFS UIFQVSQPTFPGBiQFPQMFQFSGPSNBODFw
BHHSFHBUFEEBUBTIPVMEOPUCFVTFEXIFODPOTUSVDUJOH QSPDFTTJTUPFOTVSFUIBUUIFQFPQMFJNQBDUPO
BDPOUSPMDIBSU5IFNPTUHSBOVMBSEBUBBWBJMBCMF BQSPDFTTPSQFPQMFQFSGPSNBODFJTPQUJNBM
TIPVMECFVTFE 8FTIPVMEMPPLUPTFFXIBUUIFUSVFJNQBDUPG

QFPQMFNJHIUCFPOUIFQSPDFTT0OFXBZUP *OWFTUJHBUPSi8IZ w
BQQSPBDIUIJTJTUPMPPLGPSNFBTVSFTUIBUNPSF 5FDIOJDBMTVQQPSUQFSTPOi"MMUIFSBX
EJSFDUMZTIPXUIFJNQBDUPGQFPQMFWBSJBCJMJUZPO NBUFSJBMTBSFJOTQFD TPUIFZDBOOPUCFIBWJOH
UIFQSPDFTT*UXPVMECFGBSQSFGFSBCMFUPPCTFSWF BOJNQBDUPOUIFQSPDFTTw
UIFOVNCFSPGEFWJBUJPOTXIPTFSPPUDBVTFTBSF
EVFUPHBQTJOUIFiQFPQMFQFSGPSNBODFwQSPDFTT BTJDBMMZ CFDBVTFXFIBWFOPUGVMMZ
#
4PNFPGUIFTFNBZCFEVFUPMBDLPGUSBJOJOH DIBSBDUFSJ[FEUIFJNQBDUPGUIFSBXNBUFSJBM
NJTVOEFSTUBOEJOHPGJOTUSVDUJPOT BOEPUIFS POUIFQSPDFTTOPUBOFBTZUBTLBOEBMMUIF
reasons. SBXNBUFSJBMNFBTVSFTXFEPDPOUSPMDIBSUBSF
tPercent analytical variation.*OBRVBMJUZ JOTUBUJTUJDBMDPOUSPM JUJTBTTVNFEUIBUUIFSBX
DPOUSPMMBCPSBUPSZ XFNBZGBJMUPNFBTVSF NBUFSJBMJTOPUUIFJTTVF)PXFWFS JUNBZUVSO
POFPGUIFNPTUJNQPSUBOUNFBTVSFTGSPN PVUUIBUXFXFSFTJNQMZOPUMPPLJOHBUUIFSJHIU
BDVTUPNFSTQFSTQFDUJWF UIBUJT UIFPG BTQFDUPGUIFSBXNBUFSJBMRVBMJUZ
UPUBMWBSJBUJPOJOBOBTTBZUIBUJTEVFUP tAsset management.8IFONPOJUPSJOHUIF
NFBTVSFNFOUFSSPSPSVODFSUBJOUZ5IJTJT TVDDFTTPGBOBTTFUNBOBHFNFOUQSPDFTT XF
BOJNQPSUBOUNFBTVSFCFDBVTF GPSFYBNQMF DBOMPPLBUDPTUPGNBJOUBJOJOHBTTFUT TQBSFT
UPPNVDINFBTVSFNFOUFSSPSEFHSBEFTUIF JOWFOUPSZMFWFMT BOEPUIFSWBSJBCMFT)PXFWFS
QFSGPSNBODFPGBQSPEVDUJPODPOUSPMDIBSUUIBU XFNVTUBMTPCFTVSFUIBUXFNFBTVSFUIF
is trying to detect changes in the production JNQBDUPGUIFBTTFUNBOBHFNFOUTZTUFNPO
QSPDFTT1FSGPSNJOHBTTBZTTQFFEJMZBOE UIFQSPEVDUJPOQSPDFTT5IJTDBOCFEPOFCZ
FGGJDJFOUMZNBZOPUCFFOPVHIJGUIFFSSPSPG MPPLJOHBUUIFOVNCFSPGQSPEVDUJPOJODJEFOUT
UIFNFBTVSFNFOUQSPDFTTJTMBSHF*OHFOFSBM UIBUXFSFEVFUPFRVJQNFOUJTTVFTTVDIBT
BTUBOEBSEGPSUIFNFBTVSFNFOUFSSPSJTUIF VOFYQFDUFEFRVJQNFOUGBJMVSF XSPOHNBUFSJBMT
GPMMPXJOHFRVBUJPO PGDPOTUSVDUJPO BOEPUIFSJODJEFOUT
5IFMFTTPOMFBSOFEGSPNFYBNQMFGJWFJTUIBUXF
TIPVMEOPUDPOUSPMDIBSUBNFBTVSFTJNQMZCFDBVTF
JUJTFBTZ.FBTVSFTUIBUHJWFJOTJHIUJOUPUIFQSPDFTT
TIPVMECFTFMFDUFEGPSUIFDPOUSPMDIBSU
where .JTUIFWBSJBUJPOEVFUPNFBTVSFNFOU
and 1JTUIFWBSJBUJPOEVFUPUIFQSPDFTT EXAMPLE SIX
5IJTJTXIZDPOUSPMDIBSUJOHMBCPSBUPSZBTTBZ .BOZZFBSTBHP POFPGUIFBVUIPSTIBEB
DPOUSPMTBSFTPJNQPSUBOU*GUIFDPOUSPMTCFHJO DPOWFSTBUJPOXJUIBOFX.#"HSBEVBUFXIPIBE
to display a lack of statistical control on the KVTUKPJOFEUIFJSDPNQBOZ5IFHSBEVBUFIBECFFO
DPOUSPMDIBSU JUNBZJOEJDBUFUIBUBTTVNFE exposed to the use of control charts and was
NFBTVSFNFOUVODFSUBJOUZPGUIFBTTBZJT FBHFSUPVTFUIFNJOUIFQSPDFTTGPSXIJDIIFXBT
suspect. SFTQPOTJCMF8FEJTDVTTFEXIBUXFDPVMEEP BOE*
tRaw materials management.8IFO TVHHFTUFEIFTFOENFTPNFEBUBTP*DPVMETFUVQUIF
NPOJUPSJOHUIFQFSGPSNBODFPGBSBXNBUFSJBMT DIBSUTBOEIFDPVMEUIFONBJOUBJOUIFN)PXFWFS
NBOBHFNFOUQSPDFTT XFDPOUSPMTVDIWBSJBCMFT he was enthusiastic to apply his learning and wanted
BTRVBMJUZ EFMJWFSZ QSJDF BOEPUIFSWBSJBCMFT UPTFUVQUIFDIBSUTIJNTFMG8FBHSFFEBOEBGFX
"HBJOUIFNPUJWBUJPOGPSUIJTJTUIBUUIFTF EBZTMBUFSIFTFOUNFIJTDPOUSPMDIBSUTBTQJDUVSFT
NFBTVSFTBSFFBTZUPHFUBOETIPVMECFDPOUSPM JOBEPDVNFOU5IFDIBSUTXFSFGPSEBUBTVDIBT
DIBSUFE)PXFWFS XFNBZGBJMUPNFBTVSF CBUDIZJFMET DZDMFUJNFTBOETPPO BOETPXFSF
UIFJNQBDUUIBUUIFSBXNBUFSJBMIBTPOUIF QMPUUFEPOBOJOEJWJEVBMTDPOUSPMDIBSU"TTPPOBT
NBOVGBDUVSJOHQSPDFTTXIFSFJUJTVTFE5IJTDBO *TBXUIFDIBSUT*LOFXUIFSFXBTTPNFUIJOHTVTQFDU
MFBEUPMPDBMPQUJNJ[BUJPOPGUIFSBXNBUFSJBMT BCPVUUIFDIBSUT*UBQQFBSFEUPNFUIBUUIFEBUB
NBOBHFNFOUQSPDFTTBUUIFFYQFOTFPGUIF TIPVMEJOEJDBUFPVUPGDPOUSPMTJHOBMT"MMEBUBGFMM
QFSGPSNBODFPGUIFQSPEVDUJPOQSPDFTT*UJT XFMMXJUIJOUIFDPOUSPMMJNJUTIFKVEHFEUIBUBMM
BMTPUIFSFBTPOXIZXFNBZIFBSUIFGPMMPXJOH NFBTVSFTXFSFJODPOUSPM*SFRVFTUFEUIFGJMFTXJUI
DPOWFSTBUJPOEVSJOHBSPPUDBVTFJOWFTUJHBUJPO UIFDBMDVMBUJPOT)FIBEVTFEBTQSFBETIFFUUPTFU
VQUIFDPOUSPMDIBSUT XIJDIJTGJOF
CVUJOTFUUJOH
* OWFTUJHBUPSi)BWFXFMPPLFEBUUIFSBX VQUIFDPOUSPMDIBSUMJNJUTIFIBEDBMDVMBUFEUIF
NBUFSJBMTBTUIFTPVSDFPGUIFQSPCMFN w DPOUSPMDIBSUMJNJUTBTGPMMPXT
5FDIOJDBMTVQQPSUQFSTPOi/POFFEw

Figure 3: The right and wrong way to control chart individuals data.
6 Individual Measurement of Metric
4 90
2
85 UCL=85.36
Signal
Metric
0
80 Avg=79.81
-2
-4 75 LCL=74.26
-6
4
8
12
16
20
24
28
32
36
40
44
48
0 10 20 30 40 50 60
Lot # Lot #
Signal Moving Range of Metric

90
8
7 UCL=6.82
85 Moving Range of Metric 6
5
4
80
Y
3
2 Avg=2.09
75 1
0 LCL=0.00
4
8
12
16
20
24
28
32
36
40
44
48
70
0 10 20 30 40 50 60 Lot #
Lot #
Y Metric Mean LCL UCL Correct Individuals / Moving
Range Control Chart
Incorrect Control Limits
-PXFS$POUSPM$IBSU-JNJU-$- && BCTPMVUFEJGGFSFODFCFUXFFOUIFUXPEBUBQPJOUT

6QQFS$POUSPM$IBSU-JNJU6$- && 4PJGUIFEBUBTFUIBE/EBUBQPJOUT JUXJMMIBWF/
XIFSF &JTUIFFTUJNBUFENFBOPGUIFEBUBTFU NPWJOHSBOHFWBMVFT5IFTUBOEBSEEFWJBUJPOPGUIF
and &JTUIFFTUJNBUFETUBOEBSEEFWJBUJPOPGUIF EBUBTFUDBOUIFOCFFTUJNBUFEBT
dataset.
5IFBCPWFJTBDPSSFDUBQQSPBDIUPDBMDVMBUJOH & .3
DPOUSPMDIBSUMJNJUT5IFFTUJNBUFENFBOJTTJNQMZ
UIFNFBOPGUIFEBUBTFU GPSNFECZTVNNJOHUIF XIFSF .3JTUIFBWFSBHFPGUIF/NPWJOHSBOHF
EBUBQPJOUTJOUIFEBUBTFUBOEEJWJEJOHCZUIF WBMVFT
8IFOUIFEBUBXFSFSFQMPUUFEPODPOUSPM
OVNCFSPGEBUBQPJOUTJOUIFEBUBTFU5IFFTUJNBUFE DIBSUTXJUIUIFQSPQFSDPOUSPMMJNJUT PVUPGDPOUSPM
TUBOEBSEEFWJBUJPOXBTDBMDVMBUFECZVTJOHUIF TJHOBMTXFSFDMFBS5IF.#"HSBEVBUFXIPJOJUJBMMZ
TBNQMFTUBOEBSEEFWJBUJPO*O.JDSPTPGU&YDFM DPOTUSVDUFEUIFDPOUSPMDIBSUMJNJUTVTJOHUIF
UIJTJTPCUBJOFECZVTJOHUIFTUEFWGVODUJPOBOEJT .JDSPTPGU&YDFMTUEFWGVODUJPOIBEMFBSOFEUIJT
easily calculated. This is not the correct approach. BQQSPBDIJOUIF.#"DMBTTFT5IJTJMMVTUSBUFTB
5IFQSPCMFNXJUIUIFVTFPGUIFTBNQMFTUBOEBSE DPODFSOUIBU%FNJOH
XBSOFEBCPVUNBOZZFBST
EFWJBUJPOJTUIBUJUEPFTOPUBDDPVOUGPSUIFUJNF BHPXIFOIFDBVUJPOFEBCPVUTUBUJTUJDBMNFUIPET
TFRVFODFPGUIFEBUB/PNBUUFSIPXZPVNJYBTFU CFJOHUBVHIUCZQFPQMFXIPXFSFOPUGVMMZGBNJMJBS
PGEBUB JUXJMMTUJMMIBWFUIFTBNFTBNQMFTUBOEBSE XJUIUIFN8FFOEVQXJUIUIFQIPUPDPQZFGGFDU
EFWJBUJPO4PXIBUJTSFRVJSFEJTBOFTUJNBUFPGUIF XIFSFFBDIUJNFUIFNFUIPETBOEJEFBTBSFQBTTFE
TUBOEBSEEFWJBUJPOUIBUUBLFTUIFUJNFTFRVFODF EPXO UIFZHFUGVSUIFSBOEGVSUIFSGSPNUIFPSJHJOBM
of the data into account. In practice this is done &YQFSUJTFJODPOUSPMDIBSUJOHBOEPUIFSTUBUJTUJDBM
CZVTJOHUIFNPWJOHSBOHFPGUIFEBUB'PSUXP NFUIPETPOMZDPNFTGSPNBTJHOJGJDBOUJOWFTUNFOUJO
DPOTFDVUJWFEBUBQPJOUT UIFNPWJOHSBOHFJTUIF learning and application.

"TBTJEFOPUF UIFPUIFSJTTVFXJUIUIF.#" PVSQSPDFTT UIFOUIFSFDBOCFHSFBUBEWBOUBHFTJO

HSBEVBUFTBQQSPBDIJTUIBUIFOFWFSDPOTUSVDUFEUIF QMPUUJOHUIFDPOUSPMDIBSUTJODPSSFDUMZ.PSFPGUFO
NPWJOHSBOHFDIBSU XIJDIJTBUJNFTFSJFTPGUIF UIBOOPU QMPUUJOHUIFDIBSUTVTJOHUIFJODPSSFDU
NPWJOHSBOHFWBMVFTXJUIBOBQQSPQSJBUFDPOUSPM NFUIPETEFTFOTJUJTFTUIFDIBSUTBOENBLFTPVUPG
MJNJU*UJTHFOFSBMMZBHSFFE
UIBUUIFSFJTWBMVF DPOUSPMTJHOBMTiGBEFJOUPUIFCBDLHSPVOEw)PXFWFS
JOMPPLJOHBUUIJTDIBSU CFDBVTFJUQMPUTUIFTBNF JGPVSPCKFDUJWFJTUPHBJOJOTJHIUJOUPUIFCFIBWJPSPG
EBUBJOBEJGGFSFOUXBZBOEXJMMHJWFBOPUIFSWJFX UIFQSPDFTTTPXFDBOJNQSPWFUIFQSPDFTT UIFOJUJT
of what is happening in the process. It is easy to DMFBSUIBUVTJOHUIFDPSSFDUDBMDVMBUJPOTJTJNQFSBUJWF
DPOTUSVDUXJUINPEFSODPNQVUFSTPGUXBSFBMPOH 5IFMFTTPOMFBSOFEGSPNFYBNQMFTJYJTUIBUDPOUSPM
XJUIUIFJOEJWJEVBMTDIBSU3FGFSFODFTIPXTIPX DIBSUTNVTUCFDBMDVMBUFEDPSSFDUMZ
UPDPOTUSVDUUIFNPWJOHSBOHFDIBSU
*OPSEFSUPJMMVTUSBUFUIJTQPJOU BTJNVMBUJPO CONCLUSIONS
XBTDPOTUSVDUFEXIFSFUIFCBTFEBUBDPOTJTUFE *OUIJTBSUJDMFXFIBWFFYQMPSFETPNFNPSFEPTBOE
PGBNFBTVSF:UIBUWBSJFTSBOEPNMZBDDPSEJOHUP EPOUTPGDPOUSPMDIBSUJOH5IFJNQPSUBOUMFTTPOTBSF
BOPSNBMEJTUSJCVUJPOXJUIBNFBOPGBOEB presented as follows:
TUBOEBSEEFWJBUJPOPG"EBUBTFUDPOTJTUJOHPG t%POPUVTFBHHSFHBUFEEBUBXIFODPOTUSVDUJOHB
QPJOUTXBTDSFBUFEVTJOHUIJTEJTUSJCVUJPO"TJOF control chart
XBWFJTJOUSPEVDFEPOUPUIJTCBTFTJHOBMBTTIPXOJO t%POPUDPOUSPMDIBSUBNFBTVSFTJNQMZCFDBVTFJU
UIFVQQFSMFGUQBSUPG'JHVSF5IFMPXFSMFGUDIBSU JTFBTZDPOUSPMDIBSUUIFNFBTVSFCFDBVTFJUHJWFT
TIPXTUIFDPOUSPMDIBSUVTJOHUIFTBNQMFTUBOEBSE insight into the process
EFWJBUJPOBOEUIFSJHIUIBOETJEFTIPXTUIFQSPQFS t%PNBLFTVSFUIBUUIFDPOUSPMDIBSUMJNJUTBSF
JOEJWJEVBMTBOENPWJOHSBOHFDPOUSPMDIBSUT calculated correctly.
*UJTDMFBSGSPNUIFDIBSUTUIBUUIFNFUIPEVTJOH "HBJOGPSDPOUSPMDIBSUJOH BTJOBMMFOEFBWPVST
UIFTBNQMFTUBOEBSEEFWJBUJPOUFOETUPXJEFOUIF there are the correct ways to do things and the
DPOUSPMDIBSUMJNJUTMFBEJOHUPBHSFBUFSDIBODFPG JODPSSFDUXBZT8FXJMMFYQMPSFTPNFGVSUIFSEPTBOE
NJTTJOHBOPVUPGDPOUSPMTJHOBM*UTJNQMZEPFTOPU EPOUTPGDPOUSPMDIBSUJOHJOBGVUVSFBSUJDMFJOUIJT
BDDPVOUGPSUIFDIBOHFJOUIFEBUBPWFSUJNFCFDBVTF series.
UIFSFJTOPUIJOHPGUIFUJNFTFRVFODFPGUIFEBUBJO
UIFDBMDVMBUJPOPGUIFDPOUSPMMJNJUT0OUIFPUIFS
IBOE UIFJOEJWJEVBMTDIBSUXJUIUIFQSPQFSDPOUSPM REFERENCES
MJNJUTJTNVDINPSFTFOTJUJWFUPUIFDIBOHF/PUF .D$POOFMM +PIO #SJBO,/VOOBMMZ BOE#FSOBSE
BMTPUIBUUIFNPWJOHSBOHFDIBSU TIPXOPOUIF .D(BSWFZ i%PTBOE%POUTPG$POUSPM$IBSUJOHo1BSU* w
MPXFSSJHIUIBOETJEFPG'JHVSF TIPXTOPPVUPG Journal of Validation Technology 7PM /P 8JOUFS
DPOUSPMTJHOBMT5IBUJTVTFGVMJOGPSNBUJPO CFDBVTF 8IFFMFS %+ $IBNCFST %4 Understanding Statistical
JGXFEJEOPULOPXXIBUXBTHPJOHPO JUUFMMTVT Process Control OEFEJUJPO 41$1SFTT ,OPYWJMMF
that the change in the process that led to the out of 5FOOFTTFF *4#/
DPOUSPMTJHOBMTJOUIFJOEJWJEVBMTDIBSUBSFOPUEVFUP %FNJOH 8& Out of the Crisis QVCMJTIFECZ5IF
BDIBOHFJOUIFWBSJBCJMJUZPGUIFQSPDFTT CVUNPSF $FOUFSGPS"EWBODFE&OHJOFFSJOH4UVEZ .*5
MJLFMZEVFUPBUSFOEJOUIFBWFSBHFCFIBWJPSPGUIF $BNCSJEHF .BTT *4#/JVT
data.
"TBTJEFOPUF JGPVSPCKFDUJWFJOQMPUUJOHEBUB
POBDPOUSPMDIBSUJTUPKVTUJGZUIFWBSJBCJMJUZJO
0SJHJOBMMZQVCMJTIFEJOUIF"VUVNOJTTVFPGThe Journal of Validation Technology

Pramote Cholayudth
Understanding the Process

Nonconformity ConceptThe
Scientific Reason for Process
Validation
Pramote Cholayudth
INTRODUCTION Figure 1: A process model.

6OEFSBQSPQFSDPOUSPMBOEPQUJNJ[FEDPOEJUJPO B
NBOVGBDUVSJOHQSPDFTT TFF'JHVSF
XJMMQSPEVDF Development Condition
UIFQSPEVDUDPOTJTUFOUMZNFFUJOHUIFTQFDJGJDBUJPOT Optimized
JF TQFDJGJDBUJPOGPSFBDIRVBMJUZBUUSJCVUFJT Condition
DPOTJTUFOUMZNFU
5IFTFBUUSJCVUFTOPSNBMMZIBWF
UXPMFWFMT JF DPOGPSNJOHBOEOPODPOGPSNJOH
UPUIFTQFDJGJDBUJPO
*GUIFBUUSJCVUFNFBTVSFNFOU Input Process Output
QSPDFTTJTBDDVSBUF UFTUJOHBCBUDIPGHPPERVBMJUZ
QSPEVDUXJMMSFTVMUJOGFXFSOPODPOGPSNJOHSFTVMUT
Controlled In-Process Consistent
UIBOUFTUJOHBCBUDIPGQPPSRVBMJUZ4BNQMJOHGPS Control (e.g., SPC)
Materials Quality
OPODPOGPSNJUZJTFTTFOUJBMFWFOUIPVHIFYDFMMFOU
RVBMJUZ XJUIWFSZMPXMFWFMTPGOPODPOGPSNBODF
Manufacturing Condition
DBOOPUCFBTTVSFECZOPSNBMRVBMJUZDPOUSPM 2$

TBNQMJOHQMBOT)PXFWFS UIFSFJTBSFMBUJPOTIJQ
CFUXFFOUIFUSVFOPODPOGPSNJOHSBUFPGBQSPDFTT Principles and Practices (1) with respect to the
BOEUIFQSPCBCJMJUZUIBUBTBNQMFGSPNBCBUDIXJMM TUBUJTUJDBMQSPDFTTDPOUSPM 41$
EBUBDPMMFDUJPOBOE
DPOGPSNUPUIFTQFDJGJDBUJPO JF UIFMPXFSUIFSBUF FWBMVBUJPOQMBOOJOH
UIFIJHIFSUIFQSPCBCJMJUZ
5IJTBSUJDMFEFTDSJCFT
IPXUIFPWFSBMMQSPDFTTOPODPOGPSNJOHSBUF UNDERSTANDING THE BATCH
JOGMVFODFTUIFCBUDIBOETBNQMFOPODPOGPSNJUZ NONCONFORMING RATE
BOEUIFTDJFOUJGJDSFBTPOGPSQSPDFTTWBMJEBUJPO "UZQJDBMFYBNQMFUPEFNPOTUSBUFUIF
VTJOHUIFOPODPOGPSNJUZDPODFQU*UMBZTUIF OPODPOGPSNJUZDPODFQUJTUIF6OJUFE4UBUFT
GPVOEBUJPOGPSBGVUVSFBSUJDMFUIBUXJMMSFWJFXUIF 1IBSNBDPQFJB 641
DPOUFOUVOJGPSNJUZ $6

64'PPEBOE%SVH"ENJOJTUSBUJPOTDraft UFTU 641-FUTDPOTJEFSBCFODINBSL

Guidance for IndustryProcess Validation: General CBUDIPGEPTBHFVOJUTJOXIJDIUIF$6JTOPSNBMMZ
[
ABOUT THE AUTHOR
For more Author Pramote Cholayudth is a validation consultant to Biolab, one of the top five leading pharmaceu-
information, tical manufacturers in Thailand, and executive director of Valitech, a GMP and validation service firm.
go to He is a guest speaker on process validation to the industry organized by local FDA. He was a lecturer
gxpandjvt.com/bios in a university from 1998-2001. Prior to entering the academic arena, he spent 23 years in the phar-
maceutical industry. He can be contacted by e-mail at cpramote2000@yahoo.com.

Pramote Cholayudth
Figure 2: Distribution curve for USP 29 content uniformity CZUIJTDSJUFSJPO TVDIBEPTBHFVOJUJT

standard (mean=100% LC; SD=10% LC; red area=1.24%). DMFBSMZOPODPOGPSNJOH5IFFYQFDUFE
4.5% USVFCBUDIOPODPOGPSNJOHSBUFJOUIF
4.0% GJHVSFTJTFRVBMUP JF UIF
3.5% percentage of the red area outside the
% Distribution
3.0%
2.5%
SBOHFo-$
5IFSBUFNBZCF
2.0% DBMDVMBUFEVTJOH.4&YDFMGVODUJPO
1.5% /03.4%*45

/03
1.0% 0.62% 0.62% .4%*45

0.5% 5IF$6EJTUSJCVUJPOGPSB
0.0%
0 0 0 0 0 0 0 0 0 0 00 .00 .00 .00 .00 .00 .00 .00 .00 .00 .00
NBOVGBDUVSFECBUDIXJUIBUSVFNFBO
.0 .0 .0 .0 .0 .0 .0 .0 .0 .0 0.
50 55 60 65 70 75 80 85 90 95 10
5 0 5 0 5 0 5
10 11 11 12 12 13 13 14 14 15
0 5 0
BOE34%PG-$BOE
Active content (% LA) SFTQFDUJWFMZ JTJMMVTUSBUFEJO'JHVSF
8IFOTDBMFEVQ 'JHVSFEFNPOTUSBUFT
UIBUUIFSFBSFTUJMMOPODPOGPSNJOH
Figure 3: Distribution curve for USP 30 content uniformity EBUBFNCFEEFEJOUIFCBUDIPGFWFO
standard (mean=100% LC; SD=10%; red area=1.24%). FYDFMMFOURVBMJUZQSPEVDU5IFSBUFNBZ
CFDBMDVMBUFEVTJOH.4&YDFMGVODUJPO
4.5%
4.0% /03.4%*45

/03
3.5% .4%*45

% Distribution
3.0% "TJNJMBSFYBNQMFJTUIF641
2.5% dissolution test in which a dosage unit
2.0%
XJUIBEJTTPMVUJPOPGMFTTUIBO2JT
1.5%
1.0% 0.62% 0.62% DMFBSMZOPODPOGPSNJOH XIFSF2JTUIF
0.5% product specific percent dissolution
0.0% MFWFM*GBNBOVGBDUVSFECBUDIIBTB
0 0 0 0 0 0 0 0 0 0 00 .00 .00 .00 .00 .00 .00 .00 .00 .00 .00
50
.0
55
.0
60
.0
65
.0
70
.0
75
.0
80
.0
85
.0
90
.0
95
.0 0. 5 0 5 0 5 0 5 0 5 0 USVFNFBOBOETUBOEBSEEFWJBUJPO 4%

10 10 11 11 12 12 13 13 14 14 15
Active content (% LA) PG2BOE-$ SFTQFDUJWFMZ
UIFEJTUSJCVUJPOPGEJTTPMVUJPOSFTVMUT
JTJMMVTUSBUFEJO'JHVSF5IFSBUFNBZ
CFDBMDVMBUFEVTJOH.4&YDFMGVODUJPO
Figure 4: Distribution curve for USP 30 content unifor- /03.4%*45 2
2

mity data (mean=100%; SD=6.4% LC). /03.4%*45

7.0%
6.0% UNDERSTANDING THE
BINOMIAL DISTRIBUTION
% Distribution
5.0%
8IFOTBNQMJOHEPTBHFVOJUT FH
4.0%
UBCMFUT DBQTVMFT
SBOEPNMZGSPNB
3.0%
CBUDI UIFOVNCFSPGOPODPOGPSNJOH
2.0%
VOJUTDPOUBJOFEJOUIFTBNQMFTXJMM
1.0%
GPMMPXBCJOPNJBMEJTUSJCVUJPO
0.0% #JOPNJBMEJTUSJCVUJPONBZCF
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
.0 . . . . . . . . . . . . . . . . . . . .
50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145 150 EFTDSJCFECZUIFGPMMPXJOHFYBNQMF
Active content (% LA) 8IFONBOZSBOEPNTBNQMFTPG
TJ[FOBSFUBLFOXJUISFQMBDFNFOU
EJTUSJCVUFEXJUIUSVFNFBOBOESFMBUJWFTUBOEBSE GSPNBCBUDIDPOUBJOJOHHPPE
EFWJBUJPO 34%
PGMBCFMDMBJN -$
BOE BOEOPODPOGPSNJOH FH TQFDLMFE
UBCMFUT
SFTQFDUJWFMZ5IF$6EJTUSJCVUJPOJTJMMVTUSBUFEJO UIFOVNCFSPGPCTFSWFEOPODPOGPSNJOHUBCMFUT
'JHVSFTBOEVOEFSUIFPVUEBUFE641
BOE D
BOEPCTFSWFEOPODPOGPSNBODFSBUF DO

UIFDVSSFOU641
WFSTJPOTPG6415IF XJMMWBSZBTJMMVTUSBUFEJO'JHVSFTBOE/PUF
ZFMMPXBSFBJO'JHVSFSFQSFTFOUTUIFBSFBCFUXFFO UIBUBDPOUJOVPVTEJTUSJCVUJPOJTJMMVTUSBUFEJO
BOE-$JOUIFPVUEBUFE641 UIFTFGJHVSFT CVUJUNVTUCFSFNFNCFSFEUIBU
PSFBSMJFS
DSJUFSJB'JHVSFSFQSFTFOUTUIFDSJUFSJB POMZEJTDSFUFMFWFMTPGSBUF DO
BSFQPTTJCMF
JOUIFDVSSFOU641 PSMBUFS
DSJUFSJB*OCPUI *O'JHVSF O
UIFQSPCBCJMJUZNBTTBUB
WFSTJPOTPG641 UIFSFJT[FSPUPMFSBODFGPS OPODPOGPSNBODFSBUFPG D
POUIF9BYJT
BOPCTFSWFE$6SFTVMUCFZPOEUP-$BOE NBZCFDPNQVUFE GPSFYBNQMF JO&YDFMBTGPMMPXT

Pramote Cholayudth
Figure 5: Distribution curve for USP 30 content uniformity JTUIFTBNQMFTJ[F

data (mean=100% LC; SD=6.4% LC; red area=0.01%). DPSSFTQPOETUPUIFUSVF
0.020% OPODPOGPSNJOHSBUFPG
0.018%
0.016%
'"-4&JTUIFGVODUJPODPNNBOEGPS
% Distribution
0.014% OPODVNVMBUJWFQSPCBCJMJUZGVODUJPO
0.012%
0.010%
0.008%
One can see that the range of
0.006%
0.005% 0.005% OPODPOGPSNBODFSBUFTJOTBNQMFTXJMM
0.004%
EFQFOEIFBWJMZPOTBNQMFTJ[F'PS
0.002%
0.000% JOTUBODF BUSVFOPODPOGPSNBODFSBUF
.0
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
. . . . . . . . . . . . . . . . . . . . PGJOUIFCBUDIXJMMHFOFSBUFo
50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145 150
Active content (% LA)
PCTFSWFEOPODPOGPSNBODFJOTBNQMFT
PGTJ[F TFF'JHVSF
CVUo
PCTFSWFEOPODPOGPSNBODFJOTBNQMFT
PGTJ[F5IJTJTUSVFEFTQJUFUIFGBDU
Figure 6: Distribution curve for USP 30 dissolution data UIBUUIFUSVFOPODPOGPSNBODFSBUF
(mean=Q+2.5% LC; SD=10% LC; green=95.99%; yellow=3.71%, XIJDIJTFRVBMUPUIFNFBOSBUFGPSUIF
red=0.30%). CJOPNJBMEJTUSJCVUJPO
JTUIFTBNFJO
CPUIDBTFT
4.5%
Mean = Q+2.5
4.0%
UNDERSTANDING THE
3.5%
% Distribution
PROCESS NONCONFORMING
3.0%
2.5%
RATE
2.0%
5IFQSPDFTTDPOUBNJOBUJPOSBUFJONFEJB
1.5%
GJMMT OPUNPSFUIBO JTUIFUZQJDBM
Nonconformity NPEFMGPSQSPDFTTOPODPOGPSNJOH
1.0% = 0.30%
0.5% SBUFDPODFQU'JHVSFJMMVTUSBUFTUIF
0.0% CJOPNJBMEJTUSJCVUJPOTPGUIFOVNCFSPG
Q
-5
0 45 40 35 30 25 20 15 10
Q
-
Q
-
Q
-
Q
-
Q
-
Q
-
Q
-
Q
- Q
-5 Q +5 10 15 20 25 30 35 40 45 50
Q Q + + + + + + + + + DPOUBNJOBUFEVOJUTFYQFDUFEJOSFQFBUFE
Q Q Q Q Q Q Q Q
Active content (% LA) USJBMSVOTPGUXPEJGGFSFOUNFEJBGJMM
TJ[FT GPSBOBTFQUJDGJMMJOHQSPDFTTXJUI
BUSVFDPOUBNJOBUJPO OPODPOGPSNBODF

SBUFPG5IFEFUBJMGPSNFEJBGJMMTJT
Figure 7: Distribution of nonconforming rates in samples
CFZPOEUIFTDPQFPGUIJTBSUJDMF
(batch nonconforming rate:1.0%, sample size: 300 units).
'SPNBTBNQMJOHEJTUSJCVUJPOQPJOU
25.0% PGWJFX UIFCBUDIFTQSPEVDFEBSF
iTBNQMFTwGSPNUIFQSPEVDUJPOQSPDFTT
20.0% BOEUIFJSOPODPOGPSNJOHSBUFTXJMM
Probability (%)
1.33%, 16.89% GPMMPXBCJOPNJBMEJTUSJCVUJPO*OB

15.0%
Mean = p = 0.01 = 1% CBUDIPGEPTBHFVOJUT GPSFYBNQMF UIF
10.0%
SD = (pq/n)^0.5 = 0.57% DPOUFOUVOJGPSNJUZEBUBNBZIBWFB
p = nonconforming rate = 1% = 0.01 OPODPOGPSNJUZPOBWFSBHF 'JHVSF
q = 1-p = 0.99
5.0%
n = 300

XIJMFJUTEJTTPMVUJPOEBUBNBZIBWF
BSBUFPOBWFSBHF 'JHVSF
*O
0.0% 'JHVSF UIFQSPDFTTOPODPOGPSNJOH
% % 7% 0% % % 0% 3% 7% 0% % % 0% 3% 7% 0%
00 33 6 0 33 .67 0 3 6 0 33 .67 0 3 6 0 SBUFBUXJMMHFOFSBUFUISFFEJGGFSFOU
0. 0. 0. 1. 1. 1 2. 2. 2. 3. 3. 3 4. 4. 4. 5.
Non-conforming rates in samples CBUDIOPODPOGPSNJOHSBUFEJTUSJCVUJPOT
PGCBUDITJ[FT BOE
VOJUT
FWFOUIPVHIUIFUSVF
#*/0.%*45 '"-4&
QSPDFTTSBUF NFBOPGUIFCJOPNJBMEJTUSJCVUJPO
JT
UIFTBNF
JOBMMUISFFDBTFT
8IFSF 5PJOUFHSBUFUIFPWFSBMMQJDUVSF EJTUSJCVUJPOT
JTUIFOVNCFSPGOPODPOGPSNJOHVOJUTBU PGUIFCBUDIBOETBNQMFOPODPOGPSNJUZNBZCF
MPDBUJPO
Y
JMMVTUSBUFEJO'JHVSF*OUIFGJHVSFPOFDBOTFFUIF

Pramote Cholayudth
Figure 8: Distribution of nonconforming rates in samples "TJMMVTUSBUFEJO'JHVSF BHSFBU

(batch nonconforming rate:1.0%, sample size:60 units). OVNCFSPG2$UFTUSFTVMUTXPVMECF
60.0% DPODFQUVBMMZSFRVJSFEUPPCUBJOBHPPE
FTUJNBUFPGUIFUSVF2$QBTTSBUFGPS
50.0%
BQSPDFTT JF BDDFQUBCMFPVUPG
Probability (%)
40.0% Mean = p = 0.01 = 1% 2$UFTUT

)PXFWFSJOQSBDUJDF
SD = (pq/n)^0.5 = 1.28% UIFWBMJEBUJPOUFTUSFTVMUT GSPNMBSHFS
30.0%
p = nonconforming rate = 1% = 0.01 TBNQMFTJ[FT CVUGFXFSOVNCFSTPG
20.0% q = 1-p = 0.99
n = 60 TBNQMFT
BSFVTFEUPQSFEJDUUIF2$
10.0% QBTTSBUFJOTUFBEPGBDUVBMMZQFSGPSNJOH
IVOESFETPG2$UFTUT TFF'JHVSF

0.0%
% % % % % % % % % % % % % % % % 'JHVSFJMMVTUSBUFTIPXBQSPDFTTJT
00 67 33 .00 67 .33 .0
0 7
.6 3.3
3
.0
0 7
.6 8.3
3
.0
0 7
.6 3.3
3
.0
0
0. 1. 3. 5 6. 8 10 11 1 15 16 1 20 21 2 25 WBMJEBUFEVTJOHBUMFBTUUISFFWBMJEBUJPO
Non-conforming rates in samples CBUDIFT
5PLOPXUIFUSVFOPODPOGPSNJOH
Figure 9: Distributions of contaminated units in media SBUFT CBUDIXJTFPSQSPDFTTXJTF
JU
fills (process contamination rate [PCR]: 0.1%). XPVMECFOFDFTTBSZUPUFTUFWFSZVOJUJO
following: BCBUDIPSQSPDFTTVTJOHBOBCTPMVUFMZ
Acceptance Limit = 0*; BDDVSBUFUFTU3FHSFUUBCMZ UIFPOMZ
Contamination Rate = 0x100/3000 = 0%
JOGPSNBUJPOBWBJMBCMFUPFTUJNBUFUIF
* 95% Condence that PCR 0.1%
25.0% USVFOPODPOGPSNBODFSBUFDPNFTGSPN
Acceptance Limit 2*;
Contamination Rate = 2x100/6300 = 0.03% TBNQMJOH5IFSFGPSF DPOGJSNBUJPO
20.0%
* 95% Condence that PCR 0.1% UIBUBQSPDFTTJTMPXJOOPODPOGPSNJUZ
Probability (%)
(Practically Recommended Target = 0) SFRVJSFTBEFNPOTUSBUJPOUIBUUIF

15.0% DIBODFPGDPOGPSNBODFUPUIF
MF Size: 3,000 Units
MF Size: 6,300 Units TQFDJGJDBUJPOJOFBDIWBMJEBUJPOCBUDI
10.0% JTIJHI*OTJOHMFTUBHFUFTUTJOWPMWJOH
DPOUJOVPVTNFBTVSFT FH BTTBZ

5.0% 0
BQQSPQSJBUFTUBUJTUJDBMUFDIOJRVFT FH
2
QSPDFTTDBQBCJMJUZJOEFY<$Q,>
BSF
0.0%
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
FNQMPZFEJOWBMJEBUJPOUPEFNPOTUSBUF
Number of contaminated units UIFIJHIDPOGPSNBODFSBUF*ONVMUJQMF
TUBHFUFTUT FH DPOUFOUVOJGPSNJUZ
Figure 10: Distributions of batch nonconforming rates (1% PSEJTTPMVUJPOGPSJNNFEJBUFSFMFBTF
average). Process NC rate (content uniformity): 1.0%. FYUFOEFESFMFBTF BOEEFMBZFESFMFBTF
EPTBHFGPSNT
EFNPOTUSBUJOHUIF
22.0%
IJHIQSPCBCJMJUZPGNFFUJOHUIF
Batch Size: 2,000 Units TQFDJGJDBUJPOJTBNPSFDIBMMFOHJOH
20.0%
Batch Size: 600,000 Units
18.0% Batch Size: 1,000,000 Units CVUTUJMMJNQPSUBOU PCKFDUJWFPGQSPDFTT
Probability (%)
16.0% WBMJEBUJPO
14.0%
12.0%
10.0%
A PRACTICAL APPROACH FOR
8.0% FINDING THE PROBABILITY OF
6.0% PASSING
4.0% MULTIPLE STAGE TESTS
2.0% One of the practical approaches
0.0%
UPFTUJNBUJOHUIFQSPCBCJMJUZPG
% % % % % % % % % % % % %
94 95 96 97 98 99 00 01 02 03 04 05 06 QBTTJOHUIFDPOUFOUVOJGPSNJUZPS
0. 0. 0. 0. 0. 0. 1. 1. 1. 1. 1. 1. 1.
Batch non-conforming rates dissolution test is the use of operating
DIBSBDUFSJTUJDDIBSUT4VDIDIBSUT JF
QSPCBCJMJUZDIBSUT
GPSNUIFCBTJTPG
t5IFFTUJNBUFPGCBUDIOPODPOGPSNJOHSBUFJTUIF #FSHVNTNFUIPE
XIJDISFRVJSFTUIF
BWFSBHFPGTBNQMFOPODPOGPSNJUZEBUB VTFPGTBNQMFNFBOBOE34%PS4%EBUB*O'JHVSF
t5IFFTUJNBUFPGQSPDFTTOPODPOGPSNJOHSBUFJT POFDBOTFFIPXUIFUSVF34%WBMVFJOGMVFODFT
UIFHSBOEBWFSBHF UIFQSPCBCJMJUZPGQBTTJOHUIFUISFFWFSTJPOT UIF

Pramote Cholayudth
Figure 11: Distributions of batch nonconforming rates BDDFQUBODFDSJUFSJBGPSUBCMFUTBOE

(0.1% average). Process NC rate (dissolution): 0.1%. DBQTVMFTEJGGFSJO641
PGUIF
7.0% 641DPOUFOUVOJGPSNJUZUFTU5IF
DVSWFGVSUIFTUUPUIFMFGUJTNPTU
6.0%
Batch Size: 1,000,000 Units DPOTFSWBUJWF BOEUIFDVSWFXJUIUIF
Probability (%)
5.0% TUFFQFTUTMPQFJTNPTUEJTDSJNJOBUJOH
4.0% "TTVNFUIBUUIFWBMJEBUJPODPOUFOU
3.0%
VOJGPSNJUZTBNQMFNFBOBOE34%
SFTVMUTGPSOEPTBHFVOJUTBSF
2.0% BOE-$ SFTQFDUJWFMZ"TTIPXO
1.0% JO'JHVSF UIFQSPCBCJMJUZPGQBTTJOH
UIF641UFTUJTEFUFSNJOFEUP
0.0%
CFBQQSPYJNBUFMZBTTVNJOHUIBU
0% 2% 4% 6% 8% 0% 2% 4% 6% 8% 0% 2% 4% 6% 8% 0% 2% 4% 6% 8% 0%
08 08 08 08 08 09 09 09 09 09 10 10 10 10 10 11 11 11 11 11 12 UIFSFJTOPCFUXFFOMPDBUJPOWBSJBCJMJUZ
0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
Batch non-conforming rates UIBUUIFTBNQMFJTPCUBJOFESBOEPNMZ
Figure 12: Nonconformity distributions in batches and samples.

NC Rates in Samples
Sample 1 = 0.8%
AV =
Batch # 1 Sample 2 = 0.9%
0.9%
NC Rate = 0.9% Sample 3 = 1.0%
Sample 1 = 0.9%
Process Batch # 2 AV =
Sample 2 = 1.0%
1.0%
NC Rate = 1.0%
Sample 1 = 1.0%
Batch # 3 Sample 2 = 1.1% AV =
1.1%
NC Rate:
Nonconforming Rate Average = 1% Average = 1%
Figure 13: Batch evaluation using QC test results.

QC Test Results
Test Result # 1/100
Test Result # 2/100
Test Result # 3/100
Test Result # 49/100

Test Result # 50/100 Production Batch
The high percentage (99%) of

successful QC test results
meeting the specication will
demonstrate the low non-
conforming rate.

Pramote Cholayudth
Figure 14: Batch validation using validation test results.
Validation Test Results

Test Result # 1
Test Result # 2 Validation Batch
The percentage of QC test results

meeting the specication is
Test Result # 3
predicted by computing the
probability of passing the spec. for a
sample using the validation data.
Figure 15: Process validation using validation test results.

Demonstrate at least three consecutive
Validation Test Results successful validation batches (i.e.,
demonstrate between-batch consistency).
Test Result 1
Test Result 2 PV Bat. # 1
Test Result 3 Low NC Rate
Test Result 1
PV Bat. # 2 Validated
Test Result 2
Process
Low Process NC Rate
Test Result 1
Test Result 2 PV Bat. # 3
Demonstrate the high probability of passing the specication by computation

using the validation data (i.e., demonstrate within-batch consistency).
GSPNUIFCBUDI BOEUIBUUIFEJTUSJCVUJPOPGUBCMFU PG2WBMVF5IFQSPCBCJMJUZDIBSUTGPSEFUFSNJOJOH

$6JTOPSNBM4VDIQSPCBCJMJUZDIBSUT JF DPOUFOU UIFQSPCBCBJMJUZPGQBTTJOHUIFEJTTPMVUJPOUFTUBSF
VOJGPSNJUZ
BSFBWBJMBCMFGPSEPXOMPBEBUIUUQ BWBJMBCMFGPSEPXOMPBEBUIUUQXXXJWUIPNFDPN
XXXJWUIPNFDPNKWUTQSJOHPSSFRVFTUFEGSPN +754VNNFSPSSFRVFTUFEGSPNDQSBNPUF!
DQSBNPUF!ZBIPPDPN5IFVTFPGUIFTFDIBSUTJT ZBIPPDPN5IFVTFPGUIFDIBSUTBOEUIFSFBTPO
BEESFTTFEJOEFUBJMJOSFGFSFODFOVNCFS GPSVTJOHiNFBO2wWBMVFJTBEESFTTFEJOEFUBJMJO
'JHVSFJMMVTUSBUFTIPXUIFWBSJPVTPCTFSWFE SFGFSFODFOVNCFS
4%BOEiNFBO2wWBMVFTJOGMVFODFUIFQSPCBCJMJUZ
PGQBTTJOHUIFEJTTPMVUJPOUFTU"HBJO UIFEFHSFFPG CONCLUSION
EJTDSJNJOBUJPOBOEDPOTFSWBUJTNDBOCFBTTFTTFE 5IFSFJTBSFMBUJPOTIJQCFUXFFOUIFQSPDFTTBOE
GPSFBDIDVSWF BOEUIFQSPCBCJMJUZPGQBTTJOH CBUDIRVBMJUZJOUFSNTPGUIFOPODPOGPSNJOHSBUF
UIF641UFTUDBOCFPCUBJOFEGSPNUIFPCTFSWFE 5IFMPXOPODPOGPSNJUZJOCBUDIPSQSPDFTTXJMM
WBMJEBUJPOEJTTPMVUJPO4%BOENFBO2'PS HFOFSBUFUIFIJHIDIBODFPGDPOGPSNJOHUPUIF
FYBNQMF XJUIBNFBO2BOE4%GSPNOEPTBHF TQFDJGJDBUJPOT JF BTTVSJOHUIFDPOTJTUFOURVBMJUZ

VOJUTPGBOE SFTQFDUJWFMZ 'JHVSF 5IJTBHSFFTXJUIUIF'%"ESBGUHVJEBODF

TIPXTUIBUUIFQSPCBCJMJUZPGQBTTJOHUIF641UFTU UIBUTUBUFT iFBDITUFQPGBNBOVGBDUVSJOHQSPDFTTJT
JTBQQSPYJNBUFMZ1MFBTFOPUFUIBUiNFBO2w DPOUSPMMFEUPBTTVSFUIBUUIFGJOJTIFEQSPEVDUNFFUT
WBMVF BNPSFVOJWFSTBMQBSBNFUFS JTVTFEJOTUFBE BMMEFTJHODIBSBDUFSJTUJDTBOERVBMJUZBUUSJCVUFT

Pramote Cholayudth
Figure 16: Probability of passing content uniformity tests (sample mean=100% LC, n=60, 95%
confidence level).
100%
Probability of passing CU test (%) USP 29: Capsule
90% USP 29: Tablet
USP 30
80%
70%
60%
50%
40%
30%
20%
10%
0%
0 5 0 5 0 5 0 5 0 5 0 5 0
2. 2. 3. 3. 4. 4. 5. 5. 6. 6. 7. 7. 8.
Sample RSD (%)
Figure 17: Probability of passing USP 30 content uniformity test (n=1x60=60, RSD=2.7-5.3%).
Approx. 98% Mean = 99.6, RSD = 4.5

100%
95% 2.7%
2.9%
90% 3.1%
3.3%
Probability (%)
85%
3.5%
80% 3.7%
3.9%
75%
4.1%
70% 4.3%
4.5%
65% 4.7%
60% 4.9%
5.1%
55% 5.3%
50%
.0 .4 .8 .2 .6 .0 .4 .8 .2 .6 .0 .4 .8 .2 .6 .0 .4 .8 .2 .6 .0 .4 .8 .2 .6 .0
95 95 95 96 96 97 97 97 98 98 99 99 99 100 100 101 101 101 102 102 103 103 103 104 104 105
CU sample mean (% LC)

Pramote Cholayudth
Figure 18: Probability of passing dissolution test (n=1x20=20, mean-Q=1.5-2.8).

100%
95% 1.5%
1.6%
90%
1.7%
85% 1.8%
Probability (%)
1.9%
80% 2.0%
2.1%
75%
2.2%
70% 2.3%
2.4%
65% 2.5%
2.6%
60%
2.7%
55% 2.8%
50%
0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4
0. 0. 0. 0. 0. 1. 1. 1. 1. 1. 2. 2. 2. 2. 2. 3. 3. 3. 3. 3. 4. 4. 4.
SD (% LC)
Figure 19: Probability of passing dissolution test (n=1x20=20, SD=1.4-4.0).
Approx. 99% Mean-Q = 3.8, SD = 2.8

100%
95% 1.4%
1.6%
90%
1.8%
85% 2.0%
Probability (%)
2.2%
80% 2.4%
2.6%
75%
2.8%
70% 3.0%
3.2%
65% 3.4%
3.6%
60%
3.8%
55% 4.0%
50%
9 1 3 5 7 9 1 3 5 7 9 1 3 5 7 9 1 3 5 7 9 1 3 5 7 9
0. 1. 1. 1. 1. 1. 2. 2. 2. 2. 2. 3. 3. 3. 3. 3. 4. 4. 4. 4. 4. 5. 5. 5. 5. 5.
Dissolution sample mean-Q value (% LC)

Pramote Cholayudth
JODMVEJOHTQFDJGJDBUJPOTw5IFIJHIDPOGPSNBODF +4#FSHVNBOE)-J i"DDFQUBODF-JNJUTGPSUIF/FX

SBUFJOUFSNTPG$Q,OPUMFTTUIBOJOTJOHMF *$)641$POUFOU6OJGPSNJUZ5FTU wPharmaceutical
TUBHFUFTUTPSUIFIJHIQSPCBCJMJUZPGNFFUJOHUIF Technolog y 0DUPCFS 7PM /P X X X
TQFDJGJDBUJPOJONVMUJQMFTUBHFUFTUTCBTFEPO QIBSNUFDIDPN
PQFSBUJOHDIBSBDUFSJTUJDTDVSWFT BTEFTDSJCFEIFSF $IPMBZVEUI 1 i6TFPGUIF#FSHVN.FUIPEBOE.4
DBOQSPWJEFBHPPETDJFOUJGJDSBUJPOBMFGPSKVTUJGZJOH &YDFMUP%FUFSNJOFUIF1SPCBCJMJUZPG1BTTJOHUIF641
WBMJEBUJPOTBNQMJOHQMBOTBOEBDDFQUBODFDSJUFSJB $POUFOU6OJGPSNJUZ5FTU wPharmaceutical Technology
7PMVNF /VNCFS 4FQUFNCFS X X X
ACKNOWLEDGMENT QIBSNUFDIDPN
5IFBVUIPSJTWFSZHSBUFGVMUPUIFSFWJFXFSPGUIJT $IPMBZVEUI 1 i6TJOH#FSHVNT/FX.FUIPEBOE
BSUJDMFGPSIJTWFSZIFMQGVMDPNNFOUTBOEBEEJUJPOBM .4&YDFMUP%FUFSNJOFUIF1SPCBCJMJUZPG1BTTJOHUIF
rewriting. /FX*$)641$POUFOU6OJGPSNJUZ5FTU wJournal of
Validation Technology 7PMVNF /P 'FCSVBSZ
$IPMBZVEUI 1 i&WBMVBUJPO PG 7BMJEBUJPO $POUFOU
REFERENCES 6OJGPSNJUZ5FTU3FTVMUT6TJOH-PXFS1SPCBCJMJUZ#PVOE
$FOUFSGPS%SVH&WBMVBUJPOBOE3FTFBSDI $%&3
'PPE %JTUSJCVUJPO$IBSUT wJournal of Validation Technology
BOE%SVH"ENJOJTUSBUJPO Draft Guidance for Industry: 7PMVNF /P 4QSJOH
Process Validation: General Principles and Practices $IPMBZVEUI 1 i&WBMVBUJPOPG7BMJEBUJPO%JTTPMVUJPO
/PWFNCFS 5FTU3FTVMUT6TJOH-PXFS1SPCBCJMJUZ#PVOE%JTUSJCVUJPO
( + .FSHFO i&WBMVBUJPO BOE 7BMJEBUJPO PG #MFOE $IBSUT wJournal of Validation Technology 7PMVNF /P
6OJGPSNJUZ"QQSPBDIFTGPS&TUBCMJTIJOH"DDFQUBODF 4VNNFSJVT
$SJUFSJB wJournal of Validation Technology 7PMVNF /P
'FCSVBSZ QQ ARTICLE ACRONYM LISTING
&NBJMNFTTBHF EBUFE+VOF GSPN%S-BSSZ1BVM CpK Process Capability Index
(FOFSBM1PMJDJFTBOE3FRVJSFNFOUT *OGPSNBUJPOBOE CU Content Uniformity
4UBOEBSET%FWFMPQNFOU 6OJUFE4UBUFT1IBSNBDPQFJBM FDA US Food and Drug Administration
$POWFOUJPO SFHBSEJOHUIFiPSJHJOBUJPOPG34%SFRVJSFNFOUT LC Label Claim
GPSUIF641DPOUFOUVOJGPSNJUZUFTU wXXXVTQPSH MS Microsoft
5IF6OJUFE4UBUFT1IBSNBDPQFJBM$POWFOUJPO *OD QC Quality Control
i6OJGPSNJUZPG%PTBHF6OJUT wUnited States RSD Relative Standard Deviation
Pharmacopeia UIFEJUJPO SD Standard Deviation (for a Sample)
5IF6OJUFE4UBUFT1IBSNBDPQFJBM$POWFOUJPO *OD SPC Statistical Process Control
i6OJGPSNJUZPG%PTBHF6OJUT wUnited States USP United States Pharmacopeia
Pharmacopeia,UIFEJUJPO
+4#FSHVN i$POTUSVDUJOH"DDFQUBODF-JNJUTGPS.VMUJQMF
4UBHF5FTUTwDrug Development and Industrial Pharmacy

.BSDFM%FLLFS *OD QQ

VariationPast, Present, and

Future
Welcome to Analysis and Control of Variation. KEY POINTS DISCUSSED

We have seen many approaches to improving quality, The following key points are discussed in this article:
compliance, and productivity during the past 30 years. t5IFQSJODJQMFTBOEQSBDUJDFTGPSCSJOHJOHBQSPDFTT
However, from the approaches of Shewhart, Deming, JOUPBDPOUSPMMFEBOEDBQBCMFTUBUFBOEUIFO
and Juran to TQM, Kaizen, Kanban, and Six Sigma NBJOUBJOJOHUIBUDPOEJUJPOPWFSUJNFBSFXFMM
there is a golden thread that runs through all of these EFWFMPQFE
approaches: Understanding and reducing variation. t"iHP OPHPwBUUJUVEFGPDVTJOHPOQSPDFTT
It is axiomatic that if there is no variation, there will TQFDJGJDBUJPOTJTBTJHOJGJDBOUPCTUBDMFUP
be no deviations and no out-of-specification results. JNQSPWJOHRVBMJUZ DPNQMJBODF BOEQSPEVDUJWJUZ
Some degree of variation is unavoidable; we can nearly t5IFQIBSNBDFVUJDBMJOEVTUSZTIPVMENJOJNJ[F
always reduce variation, but we can never eliminate WBSJBUJPOCFUXFFOCBUDIFT
it. Any successful approach to validating a process and t%S84IFXIBSUDSFBUFEUIFDPODFQUPGTUBUJTUJDBM
then ensuring that it remains compliant must include QSPDFTTDPOUSPMJOUIFT
elements that help us to reduce variation and then to t"RVBMJUZNBOVGBDUVSJOHQSPDFTTIBTNJOJNVN
continually monitor and control it. TUBCMFWBSJBUJPOBSPVOEUIFQSPDFTTPQUJNVN
This column is dedicated to revealing weaknesses in t5IF64'PPEBOE%SVH"ENJOJTUSBUJPOIBT
existing approaches to understanding, reducing, and SFDPNNFOEFEJEFOUJGJDBUJPOBOEDPOUSPMPG
controlling variation, and to recommend alternatives WBSJBUJPOJONBOVGBDUVSJOHQSPDFTTFTJOUIFSFDFOU
based on sound science that demonstrably work. Case QSPDFTTWBMJEBUJPOHVJEBODF
studies will be used to illustrate both problems and t"QSPDFTTJTVOEFSDPOUSPMXIFOBMMPVUDPNFT
successful methodologies. The objective of the column is GBMMBUSBOEPNBCPVUBOVODIBOHJOHNFBOBOE
to combine sound science with proven practical advice. SFNBJOXJUIJOUIFTUBUJTUJDBMMZDBMDVMBUFEDPOUSPM
Reader comments, questions, and suggestions will MJNJUTQJPOFFSFECZ4IFXIBSU
help us fulfil our objective for this column. Suggestions t5IFDPNNPOQSBDUJDFPGDBMDVMBUJOHQSPDFTTMJNJUT
for future discussion topics or questions to be addressed CZVTJOHUIFSVMFPGiNFBO PS
wJTMJLFMZ
are welcome. Case studies illustrating the successful UPNBLFFWFOUIFNPTUDIBPUJDPGEBUBMPPLTUBCMF
reduction or control of variation submitted by readers when they are not
are also most welcome. Please send your comments and t4QFDJGJDBUJPOTBSFBOFDFTTBSZQBSUPGUIFCVTJOFTT
suggestions to column coordinator John McConnell at MBOETDBQF CVUUPVTFUIFNGPSQSPDFTTBOBMZTJT
john@wysowl.com.au or journal coordinating editor BOEDPOUSPMPSGPSQSPDFTTNPOJUPSJOHJTXSPOH
Susan Haigney at shaigney@advanstar.com. t5IF4IFXIBSUBQQSPBDIJTXFMMFTUBCMJTIFE
BOETIPVMECFBQQMJFEUPQIBSNBDFVUJDBM
NBOVGBDUVSJOHQSPDFTTFT
[
ABOUT THE AUTHORS
For more Author John McConnell is the coordinator of Analysis and Control of Variation and is the owner and direc-
information, tor of Wysowl Pty Ltd. in Queensland, Australia. He may be reached at john@wysowl.com.au. Brian K.
go to Nunnally, Ph.D., is associate director, Validation, at Wyeth Research. He may be reached at nunnalb@
gxpandjvt.com/bios wyeth.com. Bernard McGarvey, Ph.D., is process modeling group leader, Process Engineering Cen-
ter, Eli Lilly and Company. He may be reached at mcgarvey_bernard@lilly.com.

INTRODUCTION TPNFUJNFTJOEJGGFSFOUDPVOUSJFT1SPEVDUJPO
5IFJNQPSUBODFPGVOEFSTUBOEJOH SFEVDJOH BOE DPTUTXPVMECFTJHOJGJDBOUMZSFEVDFE BTXPVME
DPOUSPMMJOHWBSJBUJPODBOOPUCFPWFSTUBUFE*OUPP NBJOUFOBODFDPTUT5IFDPTUPGHPPETXPVMEGBMM
NBOZQMBDFT UIFQIBSNBDFVUJDBMJOEVTUSZIBTZFU BOEUIFTUBOEBSEPGMJWJOHXPVMETPBS
to apply approaches that are routinely successful in 5IFTFXFSFTPVOEESFBNT5IFFWJEFODFUIBU
NBOZPUIFSJOEVTUSJFT3FDFOUSFHVMBUPSZEPDVNFOUT NBOVGBDUVSFSTPWFSDBNFUIFNBKPSPCTUBDMFUPNBTT
BOEQSFTFOUBUJPOTBEESFTTJOHCPUIQIBSNBDFVUJDBM QSPEVDUJPOWBSJBUJPOJTFWFSZXIFSF5IFCFOFGJUT
QSPEVDUTBOENFEJDBMEFWJDFTIBWFCFFODSJUJDBMPG CSPVHIUUPIVNBOLJOECZUIJTMBDLPGWBSJBUJPOEFGZ
UIFJOEVTUSZTBCJMJUZUPSFEVDFBOEDPOUSPMWBSJBUJPO NFBTVSFNFOU8JUIPVUNBTTQSPEVDUJPOUIFSFXPVME
BUJOQVUTBTXFMMBTJOQSPDFTT BOEKVTUJGJBCMZ CFOPUFMFWJTJPOPSPUIFSFMFDUSPOJDQSPEVDUTTVDIBT
TP5IFQSJODJQMFTBOEQSBDUJDFTGPSCSJOHJOHB DPNQVUFST OPSFGSJHFSBUPST OPQSPDFTTFEPSQBDLBHFE
QSPDFTTJOUPBDPOUSPMMFEBOEDBQBCMFTUBUFBOE GPPE BOEOPBVUPNPCJMFT5IFSFXPVMECFOPQVCMJD
UIFONBJOUBJOJOHUIBUDPOEJUJPOPWFSUJNFBSFXFMM USBOTQPSUTZTUFNT OPSFUJDVMBUJPOPGDMFBOXBUFS BOE
EFWFMPQFE)PXFWFS UIFQIBSNBDFVUJDBMJOEVTUSZ OPTFXFSBHFTZTUFNT
IBTZFUUPXJEFMZBEPQUTPNFPGUIFTFQSJODJQMFTBOE 5IFBQQMJDBUJPOPGUIJTLJOEPGUIJOLJOHUPNFEJDBM
practices. It is our intent to pursue these issues in EFWJDFTJTPCWJPVT8IBUBCPVUESVHTBOEWBDDJOFT
TVDIBXBZUIBUSFBEFSTXJMMCFBCMFUPQSPGJUGSPNUIF .PTUESVHTBOEWBDDJOFTBSFNBEFJOCBUDIFT5IF
BQQMJDBUJPOPGUIFQSJODJQMFTBOENFUIPETEJTDVTTFE TBNFQSJODJQMFTPGNBTTQSPEVDUJPOBQQMZ CVUJOB
in their daily work. TMJHIUMZEJGGFSFOUXBZ"NPMFDVMFJTEJTDPWFSFEPS
$FOUSBMUPUIJTBSUJDMFJTUIFPCTFSWBUJPOUIBU JO DSFBUFEBOEBQSPEVDUJPOQSPDFTTJTCVJMU*OQVUTBSF
NBOZDBTFT XIFOUIFUFSNTiWBSJBUJPOwBOEiQSPDFTT CMFOEFE SFBDUFE PSHSPXOBOEUIFEFTJSFENPMFDVMF
DPOUSPMwBSFVTFE BTJHOJGJDBOUQSPQPSUJPOPGQFPQMF JTDSFBUFE6TVBMMZ UIJTXJMMCFGPMMPXFECZTFWFSBM
JOUIFJOEVTUSZUSBOTMBUFUIJTUPNFBOiDPNQMJBOUw NPSFTUFQTUPDMFBWF DMFBOTF CPOE PSPUIFSXJTF
PS iSFNBJOTXJUIJOTQFDJGJDBUJPOTw PSiNFFUTUIF NPEJGZUIFNPMFDVMFVOUJMJUCFDPNFTUIFGJOBM
SFRVJSFETUBOEBSETw5IJTiHP OPHPwBUUJUVEFUPXBSET QSPEVDU5IFESVHPSWBDDJOFNBOVGBDUVSFSUSJFTUP
RVBMJUZBOEQSPDFTTDPOUSPMSFNBJOTXJEFTQSFBE BOE DMPTFMZSFQMJDBUFUIFQSPEVDU CBUDIBGUFSCBUDI)FSF
UIJTNJOETFUJTBTJHOJGJDBOUPCTUBDMFUPJNQSPWJOH UIFJOUFSDIBOHFBCJMJUZPSSFQFBUBCJMJUZPGJOUFSFTUJT
RVBMJUZ DPNQMJBODF BOEQSPEVDUJWJUZ CFUXFFOCBUDIFT5IJTJTTUJMMNBTTQSPEVDUJPO FYDFQU
UIBUUIFNBOVGBDUVSFSJTDPODFSOFEBCPVUESJWJOH
A BRIEF HISTORY OF SPECIFICATIONS EPXOUIFWBSJBUJPOCFUXFFOCBUDIFTSBUIFSUIBO
%VSJOHUIFDSBGUFSBUIBUMFEUPUIFJOEVTUSJBM CFUXFFODPNQPOFOUTJOEFWJDFT

SFWPMVUJPO DSBGUTNFONBEFBMMQSPEVDUT5IF *OMBCPSBUPSJFT UIFWBSJBUJPOPGJOUFSFTUJTGPVOE
DSBGUTNBOVTVBMMZNBEFUIFFOUJSFQSPEVDU CFUXFFOBOBMZTUTBOECFUXFFOJOTUSVNFOUT*GB
/PUJPOTTVDIBTTVCDPOUSBDUPST WFOEPST BOE CMJOEDPOUSPMTBNQMFQSPWJEFTSFQFBUBCMFSFTVMUT
outsourcing were hundreds of years away. The regardless of which analyst conducts the test or which
CBLFSHSPVOEMPDBMXIFBU DPMMFDUFEBOEQSPDFTTFE JOTUSVNFOUJTVTFE USVFJOUFSDIBOHFBCJMJUZIBTCFFO
UIFFTTFOUJBMJOHSFEJFOUT IFBUFEIJTPWFOXJUI SFBMJ[FE

XPPEHBUIFSFECZIJTGBNJMZ BOEQSPEVDFE *OUFSNTPGBEWBODJOHPVSNBUFSJBMXFMGBSFBOE
EJTUJODUJWFCSFBE#MBDLTNJUITBOEBSNPVSFST PGMFOHUIFOJOHPVSMJGFTQBO UIFBEWFOUPGNBTT
FWFONBEFUIFJSPXOSJWFUTBOEOBJMT&WFSZ QSPEVDUJPOXBTBOBXFTPNFUSJVNQI
QSPEVDUXBTPOFPGBLJOE$PNQPOFOUTXFSFOPU /FWFSUIFMFTT NBOZPGUIFJOJUJBMBUUFNQUTUPDSFBUF
JOUFSDIBOHFBCMFCFDBVTFFWFSZQBSUXBTJOEJWJEVBMMZ JOUFSDIBOHFBCMFQBSUTGBJMFE5PPPGUFOQBSUTEJEOPU
NBEFUPNBUDIPUIFSQBSUT&BDITXPSEIBEB GJUXFMMEVSJOHBTTFNCMZ SFTVMUJOHJOVOBDDFQUBCMZ
TDBCCBSETQFDJGJDBMMZDSBGUFEUPNBUDIJU*OUIF IJHITDSBQBOESFXPSL BMPOHXJUISBJTFEDPTUT
EBZTPGNVTLFUT UIFIBNNFSGSPNBCSPLFONVTLFU 5IFGJSTUCSFBLUISPVHIXBTUPSFBMJ[FUIBUFYBDU
DPVMEOPUCFSFNPWFEBOEGJUUFEUPBOPUIFSNVTLFU SFQFBUBCJMJUZXBTJNQPTTJCMF5IFTFDPOEXBT
XJUIBEBNBHFEIBNNFS*GBDPNQPOFOUGBJMFE UIFSFBMJ[BUJPOUIBUJUXBTOPUOFDFTTBSZ8IBU
BOPUIFSXBTNBEFUPPSEFSBOEIBOEGJOJTIFEUP XBTOFDFTTBSZXBTUPFOTVSFGPSUIFWBSJBCJMJUZ
NBUDIUIFPUIFSDPNQPOFOUT CFUXFFODPNQPOFOUTUPCFMPXFOPVHIUP
&BSMZJOEVTUSJBMJTUTESFBNFEPGUIF GBDJMJUBUFBTTFNCMZJOUPBXPSLBCMFQSPEVDU5IF
JOUFSDIBOHFBCJMJUZPGQBSUT BOEUIJTESFBN GJSTUQSBDUJDBMBQQSPBDIEFWFMPQFEUPBDIJFWFUIJT
DSFBUFENBOZSJDIPQQPSUVOJUJFT*GDPNQPOFOUT FOEXBTUIFDSFBUJPOPGAHPBOEAOPHPHBVHFT
XFSFJOUFSDIBOHFBCMF BXIPMFOFXDPODFQUPG *NBHJOFUIFCPSJOHPVUPGXIFFMIVCTUPBDDFQU
NBOVGBDUVSJOHXPVMECFDSFBUFE*UXPVMEBMMPX BYMFT5IFIPMFCPSFENVTUCFCJHFOPVHIUPBMMPX
QBSUTUPCFQSPEVDFEJOMBSHFWPMVNFTBOEUIFO UIFBYMFUPCFJOTFSUFE CVUOPUTPCJHUIBUUIFBYMF
BTTFNCMFEUPNBLFBQSPEVDUCZEJGGFSFOUQFPQMF XJMMGMPQBSPVOEJOUIFIVC*OUIJTDBTFBCBSPG

NFUBMXJUIPOFFOENBDIJOFEUPBTMJHIUMZTNBMMFS specification (1).

EJBNFUFSCFDPNFTUIFHBVHF5IFTNBMMFOENVTU 'SPN4IFXIBSUBOE%FNJOHUP5BHVDIJBOE4NJUI
GJUJOUPUIFXIFFMIVC CVUUIFMBSHFSFOENVTU GSPN52.BOE,BJ[FOUP,BOCBOBOE4JY4JHNB
OPU.BOVGBDUVSFSTTPPODPOWFSUFEUIFJSHP OPHP UIFSFJTBDPNNPOUIFNFUIBUSFGVTFTUPHPBXBZ
gauges into tolerances or specifications. 0VSKPCJTUPVOEFSTUBOEBOEDPORVFSWBSJBUJPO
5IJTCFDBNFUIFXBZNBOVGBDUVSFSTEFUFSNJOFE 6OGPSUVOBUFMZ NBOZJOUIFQIBSNBDFVUJDBMJOEVTUSZ
XIFUIFSUIFQBSUTNBEFDPVMECFTVDDFTTGVMMZ SFNBJOMPDLFEJOBOBODJFOUNJOETFUUIBUTBZTUIBU
BTTFNCMFEUPNBLFBTPVOEQSPEVDU5IFGJSTU QSPWJEJOHPVUDPNFTNFFUTQFDJGJDBUJPOT BMMJTXFMM
EFGJOJUJPOPGRVBMJUZXBTCPSOAll parts must remain 5IJTNJOETFUDBOCFEFNPOTUSBUFEUPCFGBMMBDJPVT
within the specifications JF JOPSPVUHPPSOPHP
BOEUPCFMJLFMZUPJODSFBTFEFWJBUJPOTBOE004
SFTVMUT5IFNPTUVQUPEBUFEFGJOJUJPOPGRVBMJUZJT
MORE RECENT DEVELOPMENTS Minimum, stable variation around the process optimum.
%FTQJUFUIFFOPSNPVTBEWBODFTNBEFEVSJOHUIF
QBTUZFBST NVDIPGUIFQIBSNBDFVUJDBMJOEVTUSZ THE CHANGING FDA THINKING
SFNBJOTMPDLFEJOUIJTHP OPHPNJOETFU*ONBOZ 5IF64'PPEBOE%SVH"ENJOJTUSBUJPOJODSFBTFEUIF
BSFBTUIFQSFWBMFOUNJOETFUTUBUFTUIBUQSPWJEJOH QSFTTVSFPOUIFQIBSNBDFVUJDBMJOEVTUSZUPCFUUFS
UIFUFTUSFTVMUVOEFSFYBNJOBUJPOSFNBJOTXJUIJO VOEFSTUBOEBOESFEVDFWBSJBUJPOXIFOJUSFMFBTFEJUT
TQFDJGJDBUJPOTOPUPOMZEPFTOPUIJOHNPSFOFFEUP ESBGUQSPDFTTWBMJEBUJPOHVJEFMJOF
UIBUTUBUFTUIF
CFTBJEBCPVUUIBUSFTVMU CVUBMTPUIBUOPUIJOHNPSF GPMMPXJOH JOQBSU
TIPVMECFTBJEBCPVUJU5IJTGVOEBNFOUBMFSSPSMFBET t2VBMJUZ TBGFUZ BOEFGGJDBDZBSFEFTJHOFEPSCVJMU
to statistically significant signals that ought to warn into the product
PGJNQFOEJOHUSPVCMFCFJOHJHOPSFE*OUVSO UIFTF t2VBMJUZDBOOPUCFBEFRVBUFMZBTTVSFENFSFMZCZ
PWFSMPPLFEPSEJTSFHBSEFETJHOBMTBOEEJTUVSCBODFT JOQSPDFTTBOEGJOJTIFEQSPEVDUJOTQFDUJPOPS
FWFOUVBMMZSFTVMUJOPVUPGTQFDJGJDBUJPO 004
SFTVMUT testing
UIBUTIPVMEBOEDPVMEIBWFCFFOBWPJEFE
5IFZ t&BDITUFQPGBNBOVGBDUVSJOHQSPDFTTJTDPOUSPMMFE
BMTPMFBEUPBOVOOFDFTTBSZJODSFBTFJOWBSJBUJPO UPBTTVSFUIBUUIFGJOJTIFEQSPEVDUNFFUTBMM
UIBU JOBDDPSEBODFXJUI-JUUMFT-BX EBNBHFTOPU EFTJHODIBSBDUFSJTUJDTBOERVBMJUZBUUSJCVUFT
POMZRVBMJUZ CVUBMTPQSPEVDUJWJUZ"OZBQQSPBDI including specifications.
UIBUCMJOETQFPQMFUPTUBUJTUJDBMMZWBMJETJHOBMTBOE
warnings is poor science. "OPUIFSTFDUJPOPGUIJT'%"HVJEFMJOF
TUBUFTUIBU
*OUIFT%S84IFXIBSUDSFBUFEUIFDPODFQU NBOVGBDUVSFSTTIPVMEEPUIFGPMMPXJOH
PGTUBUJTUJDBMDPOUSPM PSTUBCJMJUZ)FXBTBCMFUP t6OEFSTUBOEUIFTPVSDFTPGWBSJBUJPO
EFNPOTUSBUFUIBUBQSPDFTTIBEOPLOPXODBQBCJMJUZ t%FUFDUUIFQSFTFODFBOEEFHSFFPGWBSJBUJPO
VOUJMJUXBTJOTUBUJTUJDBMDPOUSPM.PSFUIBO t6OEFSTUBOEUIFJNQBDUPGWBSJBUJPOPOUIFQSPDFTT
ZFBSTMBUFSXFGJOENBOZDIFNJTUT CJPMPHJTUT BOEVMUJNBUFMZPOQSPEVDUBUUSJCVUFT
BOEFOHJOFFSTDBMDVMBUJOHQSPDFTTDBQBCJMJUZ t$POUSPMUIFWBSJBUJPOJOBNBOOFSDPNNFOTVSBUF
EFTQJUFUIFGBDUUIBUUIFEBUBBSFEFNPOTUSBCMZ with the risk it represents to the process and
VOTUBCMF"OVNCFSXJMMCFDBMDVMBUFE CVUJOUIFTF product.
DJSDVNTUBODFTJUJTMJLFMZUPCFNFBOJOHMFTT

*O %S5BHVDIJXPOUIF%FNJOH.FEBMGPS 6OGPSUVOBUFMZ UIFSFJTOPTIPSUBHFPGTDJFOUJTUT
BOBQQSPBDIXFOPXDBMMUIF5BHVDIJ-PTT'VODUJPO JOUIFJOEVTUSZXIPCFMJFWFUIBUUIFZNFFUUIFTF
&TTFOUJBMMZ 5BHVDIJDSFBUFEOFXTUBUJTUJDBMUPPMTUIBU SFRVJSFNFOUT XIFOUIFZEPOPU
EFNPOTUSBUFEUIBUFWFOXIFOBMMPVUDPNFTNFFU 5IF'%"HVJEFMJOFGVSUIFSTUBUFT
i"GUFS
TQFDJGJDBUJPOT MPTTJTJODVSSFE
)JTNFUIPET FTUBCMJTIJOHBOEDPOGJSNJOHUIFQSPDFTT
TVQQPSUFE4IFXIBSUBOE%FNJOHXIPIBEDMBJNFE NBOVGBDUVSFSTNVTUNBJOUBJOUIFQSPDFTTJOB
UIBUSFEVDJOHWBSJBUJPOXBTBMXBZTBTPVOEFDPOPNJD TUBUFPGDPOUSPMPWFSUIFMJGFPGUIFQSPDFTT FWFOBT
BQQSPBDI FWFOUIPVHIBMMEBUBXFSFXJUIJO NBUFSJBMT FRVJQNFOU QSPEVDUJPOFOWJSPONFOU
TQFDJGJDBUJPOT
QFSTPOOFM BOENBOVGBDUVSJOHQSPDFEVSFTDIBOHFw
*OUIFFBSMZTBOFOHJOFFSGSPN.PUPSPMB .S 5IJTCFHTUIFRVFTUJPO IPXTIPVMEXFEFGJOFUIF
#JMM4NJUI DBNFUPUIFTBNFDPODMVTJPOT4NJUI TUBUFNFOUiNBJOUBJOUIFQSPDFTTJOBTUBUFPGDPOUSPMw
VMUJNBUFMZMBJEGPVOEBUJPOTGPSUIFDPODFQUPG4JY %VSJOHUIF.BSDI1JUUDPODPOGFSFODFUIF
4JHNB4NJUITUBUFEUIBUBMMDSJUJDBMUPRVBMJUZ $52
PQQPSUVOJUZXBTUBLFOUPBTLNBOZTDJFOUJTUTXPSLJOH
DIBSBDUFSJTUJDTTIPVMEPDDVQZOPNPSFUIBOIBMG JOUIFQIBSNBDFVUJDBMJOEVTUSZUPQSPWJEFBEFGJOJUJPO
UIFTQFDJGJDBUJPOSBOHF*ONPSFTQFDJGJDUFSNTIF PGUIJTUFSN5IFNBKPSJUZSFTQPOEFECZTUBUJOH
TUBUFEUIBUUIFOFBSFTUTQFDJGJDBUJPOGPSBOZ$52 that a state of control existed if all results fell within
DIBSBDUFSJTUJDNVTUCFBUMFBTUGSPNUIFOFBSFTU specifications.

Figure 1: Two approaches to displaying variation.

Batch release results against specications
550
500
450
400
350
300
250
200
150
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77
Batch release data as a Shewhart control chart

550
500
450
400
350
300
250
200
150
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
5IJTJTOPUDPSSFDU"TUBUFPGDPOUSPMFYJTUTXIFOBMM *OUIJTQBSBHSBQI'%"TUBUFTDMFBSMZXIBUJT
PVUDPNFTGBMMBUSBOEPNBCPVUBOVODIBOHJOHNFBO SFRVJSFE5IFJOEVTUSZOFFETUPVTFTUBUJTUJDBMQSPDFTT
BOESFNBJOXJUIJOUIFTUBUJTUJDBMMZDBMDVMBUFEDPOUSPM DPOUSPM 41$
UFDIOJRVFTUPFWBMVBUFCPUIQSPDFTT
MJNJUTQJPOFFSFECZ4IFXIBSU TUBCJMJUZBOEQSPDFTTDBQBCJMJUZ5IJTSFRVJSFTUIFVTF
*UJTXPSUIZUPOPUFUIBUUIFDPNNPOQSBDUJDFPG PG4IFXIBSUDPOUSPMDIBSUTUPEFUFSNJOFXIFUIFSPS
DBMDVMBUJOHUIFMJNJUTPGBTZTUFNCZVTJOHUIFSVMFPG OPUBQSPDFTTJTJOBDPOUSPMMFETUBUFCVUXIZCFTP
iNFBO PS
wJTMJLFMZUPNBLFFWFOUIFNPTU QSFTDSJQUJWF
DIBPUJDPGEBUBMPPLTUBCMF5IJTNFUIPEPGDBMDVMBUJOH
MJNJUTBTTVNFTUIBUUIFEBUBBSFDPOUSPMMFEPSTUBCMF SHREWHART CONTROL CHARTS
8IFOUIFZBSFOPU VQTFUTXIJDIJODSFBTFWBSJBUJPO "TJHOJGJDBOUQBSUPGUIFBOTXFSDBOCFTFFOJOUIF
XJMMDBVTFMJNJUTDBMDVMBUFEJOTVDIBXBZUPXJEFO FYBNQMFJO'JHVSF
5XPDIBSUTGPSUIFTBNF$52
BOEUPNBLFUIFEBUBMPPLTUBCMF FWFOXIFOUIFZBSF characteristic are displayed. The top chart shows
not. SFTVMUTDPNQBSFEXJUITQFDJGJDBUJPOT BOEUIFMPXFS
'JOBMMZ UIF'%"HVJEFMJOFTUBUFT
DIBSUVTFTMJNJUTDBMDVMBUFEVTJOH4IFXIBSUTDPOUSPM
i8FSFDPNNFOEUIBUBTUBUJTUJDJBOPSQFSTPO DIBSUJOHUFDIOJRVFT5IF.PWJOH3BOHFTDIBSUJT
XJUIBEFRVBUFUSBJOJOHJOTUBUJTUJDBMQSPDFTT PNJUUFEGSPNUIF4IFXIBSUDIBSUGPSTJNQMJDJUZBOE
DPOUSPMUFDIOJRVFTEFWFMPQUIFEBUBDPMMFDUJPO clarity.
QMBOBOETUBUJTUJDBMNFUIPETBOEQSPDFEVSFTVTFE *O'JHVSF UXPSFTVMUTBSFSFQPSUFE004POF
JONFBTVSJOHBOEFWBMVBUJOHQSPDFTTTUBCJMJUZBOE BUCBUDIBOEPOFBUCBUDI#FDBVTFJUJTBO
QSPDFTTDBQBCJMJUZ1SPDFEVSFTTIPVMEEFTDSJCF JTPMBUFETQFDJBMDBVTFFWFOU JUXBTDPSSFDUUPTUVEZ
IPXUSFOEJOHBOEDBMDVMBUJPOTBSFUPCFQFSGPSNFE CBUDIJOBOBUUFNQUUPEFUFSNJOFDBVTBMJUZ5P
1SPDFEVSFTTIPVMEHVBSEBHBJOTUPWFSSFBDUJPOUP UBLFUIFTBNFBDUJPOGPSCBUDIJTBNJTUBLF
JOEJWJEVBMFWFOUTBTXFMMBTBHBJOTUGBJMVSFUPEFUFDU 5IF4IFXIBSUDPOUSPMDIBSUTIPXTBTUBUJTUJDBMMZ
QSPDFTTESJGU1SPEVDUJPOEBUBTIPVMECFDPMMFDUFE TJHOJGJDBOUVQXBSETIJGUJOUIFNFBOBUCBUDI
UPFWBMVBUFQSPDFTTTUBCJMJUZBOEDBQBCJMJUZ5IF That is a signal to which the scientists should react.
RVBMJUZVOJUTIPVMESFWJFXUIJTJOGPSNBUJPO*G #BUDISFQPSUFEBT004OPUCFDBVTFBOZUIJOH
EPOFQSPQFSMZ UIFTFFGGPSUTDBOJEFOUJGZWBSJBCJMJUZ DIBOHFEEVSJOHUIFQSPEVDUJPOPGUIBUCBUDI*UJT
JOUIFQSPDFTTBOEPSQSPEVDUUIJTJOGPSNBUJPODBO BSBOEPNQPJOUJOBTZTUFNPGEBUBUIBUTIJGUFEBU
CFVTFEUPBMFSUUIFNBOVGBDUVSFSUIBUUIFQSPDFTT CBUDI XIJDIJTXIFSFUIFSFBMSPPUDBVTFTPGUIF
TIPVMECFJNQSPWFEw TFDPOE004FWFOUXJMMCFGPVOE/FWFSUIFMFTT JO
NPTUDPNQBOJFTXIFOTJNJMBSDJSDVNTUBODFTFYJTU

Figure 2: Two approaches to calculating limits.

Limits set at mean 3
40
35
30
25
20
15
10
5
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 106 111
Data with Shewhart control limits

40
35
1-Point Average
30
25
20
15
10
5
0
1 21 41 61 81 101
XFGJOETDJFOUJTUTTUVEZJOHUIFGBJMFECBUDI SBUIFS Analytical Processes

UIBOUIFQPJOUJOUJNFXIFSFUIFQSPDFTTDIBOHFE "OZTUBUJTUJDBMBQQSPBDIDBOCFOPCFUUFSUIBOUIF
"TJNJMBSQSPCMFNDBOCFGPVOEXIFOTDJFOUJTUT EBUBJUVTFT*NBHJOFBQSPEVDUJPOQSPDFTTUIBUJT
BSFTUJMMVTJOHUIFPMEiNFBO PS
wBQQSPBDI JOTUBUJTUJDBMDPOUSPM CVUUIFBOBMZUJDBMQSPDFTTVTFE
UPDBMDVMBUFMJNJUT XIJDIJTJMMVTUSBUFEJO'JHVSF UPNFBTVSFJUJTVOTUBCMF*UJTQPTTJCMFUIBUUIFEBUB
"HBJO UXPDIBSUTBSFTIPXO#PUIDIBSUTTIPX XJMMFYIJCJUJOTUBCJMJUZ5IJTJOUVSOXJMMDBVTFBO
B$52DIBSBDUFSJTUJDJOBDIFNJDBMDMFBWFQSPDFTT JOWFTUJHBUJPOGPSDBVTFTPGFYDFTTJWFWBSJBUJPOJOUIF
GPSBCJPMPHJD5IFUPQDIBSUVTFTUIFiNFBOw QSPEVDUJPOBSFBBIVOUUIBUJTBMMXBTUFEFGGPSU
BQQSPBDI5IFMPXFSDIBSUJTB4IFXIBSUDPOUSPMDIBSU BOEUIBUJTEPPNFEUPGBJMCFDBVTFUIFQSPCMFNTBSF
BHBJO UIF.PWJOH3BOHFDIBSUJTPNJUUFE
JOUIFBOBMZUJDBMBSFB5IJTTVCKFDUBMPOFXPVMEGJMM
5PPNBOZTDJFOUJTUTDPODMVEFUIBUUIFDIBSUVTJOH TFWFSBMWPMVNFT BOEJUJTFBSNBSLFEBTUIFTVCKFDUGPS
UIFiNFBOwBQQSPBDITIPXTOPTJHOBMT*GGBDU BGVUVSFUPQJDGPSUIJTDPMVNO'PSOPXJUXJMMTVGGJDF
UIFSFBSFTFWFOTJHOBMTJOUIJTEBUBTFU BMMPGXIJDI UPJMMVTUSBUFBTJNJMBSBOBMZUJDBMJTTVFXJUIUIFUXP
BSFFYQPTFEJOUIF4IFXIBSUDPOUSPMDIBSU5IJT DIBSUTJO'JHVSF5IFUPQDIBSUTIPXTUIFMBCPSBUPSZ
JTXIBUUIF'%"HVJEFMJOFTBSFOPXEFNBOEJOH DPOUSPMTXJUIMJNJUTQMBDFEBUNFBO and the lower
UIBUTDJFOUJTUTDBOCFUUFSUFMMUIFEJGGFSFODFCFUXFFO DIBSUTIPXTUIFMBCPSBUPSZDPOUSPMTBTB4IFXIBSU
SBOEPNBOEOPOSBOEPNEBUBUIBUUIFZLOPX control chart. The hunt for the trend factory people
XIFOUPUBLFBDUJPO BOEXIFOOPUUP"DDPSEJOH TBXJOUIFQSPEVDUJPOEBUBXBTOFWFSHPJOHUPSFWFBM
UPUIF'%"HVJEFMJOFT UIFTJUVBUJPOJO'JHVSFJT SPPUDBVTFT CFDBVTFUIFUSFOEXBTJOUIFMBCPSBUPSZT
VOBDDFQUBCMF FWFOUIPVHIBMMSFTVMUTUPEBUFSFNBJOFE analytical processes.
XJUIJOTQFDJGJDBUJPOT#FDBVTFUIF'JHVSFQSPDFTTJT
OPUTUBCMFPSDPOUSPMMFE JUIBTOPLOPXODBQBCJMJUZ CONCLUSIONS
4PPOFSPSMBUFSJUXJMMHFOFSBUFBO004SFTVMUCVU *OPSEFSUPDPNQMZXJUIUIF'%"HVJEFMJOFT BOE
XIZXBJU *GUIFNBOBHFSTBOETDJFOUJTUTXPSLJOHJO UPCFUUFSHVBSBOUFFRVBMJUZBOEQSPEVDUJWJUZ
UIJTQSPDFTTEJEOPUIJOHNPSFUIBOUPTUBCJMJ[FUIF DPNQBOJFTJOUIFQIBSNBDFVUJDBMJOEVTUSZIBWFOP
QSPDFTT UIFZXPVMEBMMCVUHVBSBOUFFJUTDBQBCJMJUZ DIPJDFPUIFSUIBOUPVTF41$UFDIOJRVFTUPCSJOH
/PDIBOHFTUPDVSSFOUHPPENBOVGBDUVSJOHQSBDUJDFT BMMUIFJSDMJOJDBMUSJBMT EFWFMPQNFOU BOBMZUJDBM
BSFSFRVJSFE*OUIJTDBTF BMMUIBUJTOFDFTTBSZUP and production processes into a state of statistical
FOTVSFHPPERVBMJUZBOEQSPEVDUJWJUZJTUPTUBCJMJ[F DPOUSPM5IFHVJEFMJOFTTFFNUPMFBWFMJUUMFSPPN
UIFQSPDFTTUPNBLFJUSFQFBUBCMFBOEQSFEJDUBCMF GPSJOUFSQSFUBUJPO BOEVOEFSTUBOEBCMZTP#FDBVTF
0ODFUIFWBSJBUJPOJTSFEVDFE UIFOJUXJMMCFFBTJFSUP BOVOTUBCMFQSPDFTTIBTOPLOPXODBQBCJMJUZ XF
TFFUIFDBVTBMSFMBUJPOTIJQTBOEVMUJNBUFMZJNQSPWF DBOOPUDBMDVMBUFUIFQSPCBCJMJUZPGFJUIFSQBTTJOH
the process. PSGBJMJOHTQFDJGJDBUJPOTJOBTUBUJTUJDBMMZWBMJEXBZ

Figure 3: Trending laboratory controls.

Limits set at mean 3
37
32
27
22
17
12
4
4
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
-0
il
il
il
il
il
il
il
ay
ay
ay
ay
pr
pr
pr
pr
pr
pr
pr
ay
ay
ay
-M
-M
-M
-M
-A
-A
-A
-A
-A
-A
-A
M
11
14
17
20
23
26
29
11
14
17
20
2-
5-
8-
Data with Shewhart control limits
35
1-Point Average
30
25
20
15
04
04
04
04
04
04
04
4
4
-0
-0
-0
-0
il-
il-
il-
il-
il-
il-
il-
-0
-0
-0
ay
ay
ay
ay
pr
pr
pr
pr
pr
pr
pr
ay
ay
ay
-M
-M
-M
-M
-A
-A
-A
-A
-A
-A
-A
M
11
14
17
20
23
26
29
11
14
17
20
2-
5-
8-
VOUJMUIFQSPDFTTVOEFSFYBNJOBUJPOJTLOPXOUP 8"4IFXIBSU Economic Control of Quality of Manufactured

CFTUBCMF
0VSDPOTVNFSTIBWFBSJHIUUPFYQFDU Product 7BO/PTUSBOE
UIBUUIFNFEJDJOFTUIFZUBLFDPNFGSPNDPOUSPMMFE 8"4IFXIBSU Statistical Method from the Viewpoint of
BOEQSFEJDUBCMFQSPDFTTFT0VSTUPDLIPMEFST Quality Control 5IF(SBEVBUF4DIPPMPG"HSJDVMUVSF
IBWFBSJHIUUPFYQFDUUIBUXFXJMMOPUUISPXBXBZ 8BTIJOHUPO
UIFJSJOWFTUNFOUXJUIVOOFDFTTBSZWBSJBUJPOBOE 884DIFSLFOCBDI The Deming Route to Quality and
EFWJBUJPOT0VSFNQMPZFFTIBWFBSJHIUUPUBLFQSJEF Productivity $&&1SFTT
BOEKPZJOUIFJSXPSL1SPEVDUJPOQFPQMFIBWFB 8&%FNJOH Out of the Crisis .*5
SJHIUUPFYQFDUUIBUUIFBOBMZUJDBMTZTUFNTPOXIJDI '%" (VJEBODFGPS*OEVTUSZ Process Validation: General
UIFZBSFTPEFQFOEFOUBSFTUBCMF#SJOHJOHPVS Principles and PracticesDraft Guidance JVT
processes into statistical control will help us to satisfy
all these needs. ARTICLE ACRONYM LISTING
4QFDJGJDBUJPOTBSFBOFDFTTBSZQBSUPGUIFCVTJOFTT CTQ Critical to Quality
MBOETDBQF CVUUPVTFUIFNGPSQSPDFTTBOBMZTJTBOE FDA US Food and Drug Administration
DPOUSPMPSGPSQSPDFTTNPOJUPSJOHJTBSFDJQFGPS OOS Out-Of-Specification
USPVCMF5IF4IFXIBSUBQQSPBDIJTJEFBMGPSUIFTF SPC Statistical Process Control
purposes. It is a tried and true approach that has
TUPPEUIFUFTUPGUJNF"MMUIBUSFNBJOTJTGPSUIF
QIBSNBDFVUJDBMJOEVTUSZUPVTFJUXJEFMZBOEXJTFMZ
REFERENCES
#,/VOOBMMZBOE+4.D$POOFMM Six Sigma in the
Pharmaceutical Industry $3$1SFTT
#,/VOOBMMZ %'/VOOBMMZ BOE+4.D$POOFMM
i"WPJEJOHUIF1BJOPG0VUPG4QFDJGJDBUJPO3FTVMUT w
BioPharm International +VOF

Having It All
Analysis and Control of Variation is dedicated INTRODUCTION AND HYPOTHESIS

to revealing weaknesses in existing approaches to It is ironic that to this day we find people engaged
understanding, reducing, and controlling variation JOBRVBMJUZWFSTVTRVBOUJUZEFCBUF4PNFNBOBHFST
and to recommend alternatives that are not only BSFDPODFSOFEUIBUUIFJOTUBMMNFOUPGJNQSPWFE
based on sound science, but also which demonstrably RVBMJUZQSPDFEVSFTXJMMSFTVMUJOTMPXFSQSPEVDUJPO
work. The objective of the column is to combine and higher unit costs.
sound science with proven practical advice. The central hypothesis of this article is that
Reader comments, questions, and suggestions will XIFOUIFBQQSPBDIUPJNQSPWFERVBMJUZCBTFEPO
help us fulfill our objective for this column. Please SFEVDJOHWBSJBUJPOJTTVDDFTTGVMMZVTFE JUMFBETUP
send your comments and suggestions to column IJHIFSUISPVHIQVUWPMVNFDBQBDJUZ MPXFSXPSL
coordinator John McConnell at john@wysowl.com. JOQSPHSFTTBOEDZDMFUJNF SFEVDFEVOJUDPTUT
au or journal coordinating editor Susan Haigney at JNQSPWFEDPNQMJBODF BOEJNQSPWFEDVTUPNFS
shaigney@advanstar.com. TFSWJDFMFWFMT8FDBOIBWFJUBMM JGXFLOPXIPXUP
SFEVDFWBSJBUJPO
KEY POINTS DISCUSSED
The following key points are discussed in this TYPES OF VARIATION
article: 'PSUIFQVSQPTFTPGUIJTBSUJDMF WBSJBUJPODBOCF
t5IFSFJTOPUSBEFPGGCFUXFFORVBMJUZ CSPLFOJOUPUISFFDBUFHPSJFTWBSJBUJPOJOQSPDFTT
QSPEVDUJWJUZ RVBOUJUZ
BOEDPNQMJBODF PSQSPDFEVSF WBSJBUJPOJORVBMJUZDIBSBDUFSJTUJD
t4PNFTJNQMFTUBUJTUJDBMDPODFQUTEFNPOTUSBUF FH Q) UFNQFSBUVSF TQFDJGJDBDUJWJUZ BOEQVSJUZ

that only operations that successfully reduce BOEWBSJBUJPOJOWPMVNF JF UIFEBJMZ XFFLMZ PS
WBSJBUJPOBOEJNQSPWFRVBMJUZXJMMBDIJFWF NPOUIMZQSPEVDUJPOWPMVNF
"TXJMMCFTFFO UIFSF
IJHIPVUQVU MPXXPSLJOQSPHSFTT BOEHPPE BSFMJOLTCFUXFFOUIFTFUISFFUZQFT CVUGPSFBTFPG
DPNQMJBODF EJTDVTTJPOXFDBOUSFBUUIFNTFQBSBUFMZ
t5IFSFBSFTFWFSBMUZQFTPGWBSJBUJPO CVUVTVBMMZ
BMMBSFJOUFSMJOLFEUPTPNFEFHSFF VARIATION IN PROCESS AND
t7BSJBUJPOJTUIFDIJFGFOFNZPGRVBMJUZ PROCEDURE
QSPEVDUJPO BOEDPNQMJBODF "NJTUBLFJTNBEFJOUIFMBCPSBUPSZ5IFBOBMZTU
t-JUUMFT-BXFYQMBJOTIPXVOEFSTUBOEJOHBOE NBEFBOFSSPSJOUIFTBNQMFQSFQBSBUJPOTQFDJGJFE
SFEVDJOHWBSJBUJPOJODSFBTFTWPMVNFPVUQVU CZUIFQSPDFEVSF BHPPENBOVGBDUVSJOHQSBDUJDF
t1PPSRVBMJUZBOEFSSBUJDQSPDFTTFTBOE (.1
WJPMBUJPO-BUFSUIBUEBZBOPVUPG
QSPDFEVSFTBMMJODSFBTFWBSJBUJPO TQFDJGJDBUJPO 004
SFTVMUJTSFDPSEFE"GFXIPVST
MBUFSBQMBOUPQFSBUPSNBLFTBOFSSPSJOBCBUDI
TFUVQ5IFCBUDIJTTDSBQQFE
[
ABOUT THE AUTHORS
For more Author John McConnell is the coordinator of Analysis and Control of Variation and is the owner and
information, director of Wysowl Pty Ltd. in Queensland, Australia. He may be reached at john@wysowl.com.au.
go to Brian K. Nunnally, Ph.D., is associate director, Validation, at Wyeth Research. He may be reached
gxpandjvt.com/bios at nunnalb@wyeth.com. Bernard McGarvey, Ph.D., is process modeling group leader, Process
Engineering Center, Eli Lilly and Company. He may be reached at mcgarvey_bernard@lilly.com.

&WFSZEBZEFWJBUJPOTBSFSBJTFEBOEJOWFTUJHBUJPOT BSFSBSF*OBDDPSEBODFXJUI+VSBOTSVMF BCPVU

BSFDPOEVDUFE*OUIFBCPWFDBTFT UIFRVBSBOUJOFE PGUIFEFWJBUJPOTSJTFGSPNBCPVU
CBUDIJTDMFBSFEGPSSFMFBTFXIFOJUJTEFUFSNJOFE PGUIFDBVTFT5IFKPCJTOPUBTCJHBTJUTPNFUJNFT
UIBUBOBOBMZUJDBMFSSPSXBTUIFQSPCMFNBOEUIBU TFFNT*OPOFMBCPSBUPSZ EFWJBUJPOTESPQQFEUP
the product was fine. On the production line a MFTTUIBOIBMGUIFJSQSFWJPVTMFWFMJOBNBUUFSPG
OFXCBUDIXBTSVTIFEJOUPQSPEVDUJPOBGUFSUIFCBE weeks.
CBUDIXBTDMFBSFEGSPNUIFQSPDFTT*OWFTUJHBUJPOT
XFSFDPNQMFUFE BOEUIFEFWJBUJPOTXFSFDMFBSFE Variation Is The Cause
5IFBOBMZUJDBMBOEQSPEVDUJPOQSPDFTTFTSFNBJO *UJTBYJPNBUJDUIBUEFWJBUJPOTBSFDBVTFECZ
DPNQMJBOUBOEESVHTGMPXUPQBUJFOUT WBSJBUJPO5IFBOBMZTUWBSJFTGSPNUIFQSPDFEVSF
5IFBVUIPSTIBWFCFFOUFMMJOHUIJTTUPSZBU JOTFMFDUJOHBCVGGFS"OPQFSBUPSWBSJFTGSPNUIF
TFNJOBSTBOEDPOGFSFODFTGPSUIFQIBSNBDFVUJDBM TUBOEBSE CVUQFSIBQTBNCJHVPVT (.1TFUVQ
JOEVTUSZGPSZFBST.BOZJOUIFBVEJFODFTBSF QSPDFEVSF5IFSFJTNJOVUFUPNJOVUFBOECBUDI
RVJ[[JDBM5IFTUPSZTFFNTTPPCWJPVT%FWJBUJPOT UPCBUDIWBSJBUJPOJOFWFSZQSPDFTT)PXFWFS NVDI
IBQQFO5IFEFWJBUJPOTBSFDMFBSFE BOEQSPEVDUJPO PGUIJTWBSJBUJPOJTJHOPSFEGPSBTMPOHBTJUSFNBJOT
DPOUJOVFT8IBUJTUIFNPSBMPGUIFTFDBTFTUVEJFT within specifications. To ignore trends and other
TJHOBMTVOUJMTQFDJGJDBUJPOTBSFCSFBDIFEJTUIFTIPSU
Dr. Demings Story SPBEUP)BEFT
%VSJOHIJTTFNJOBST%S8&EXBSET%FNJOHUPMEB *GUIFBQQSPBDIUPEFWJBUJPOTVTFECZTPNVDIPG
TUPSZBCPVUBTUVEFOUIFXPVMEQJDLVQPOIJTXBZ UIFQIBSNBDFVUJDBMJOEVTUSZBDUVBMMZXPSLFE BOZ
UPDMBTTFTBU/FX:PSL6OJWFSTJUZ5IFTUVEFOUIBE QMBOUTIPVMETFFBEFDMJOFJOUIFUPUBMOVNCFSPG
BTPOPGBCPVUGPVSPSGJWFZFBSTPGBHF0OPOF EFWJBUJPOTBTUIFNPOUITBOEZFBSTQBTT*ONBOZ
PDDBTJPO%S%FNJOHTQBTTFOHFSXBTSVOOJOHMBUF DBTFT UIFZEPOPU5IFDIBSUJO'JHVSFDPNFT
5IFTPOJOWJUFE%S%FNJOHJOTJEFBOEPGGFSFEIJN GSPNBMBSHFQIBSNBDFVUJDBMTJUF0OMZEFWJBUJPOT
UPBTU%FNJOHRVFSJFEXIFUIFSUIFCPZXBTPME GSPNQMBOUTBOEMBCPSBUPSJFTJOPQFSBUJPOGPSUIF
FOPVHIUPGJMMTVDIBOPSEFS5IFCPZXBTPGGFOEFE FOUJSFGPVSZFBSTDPWFSFECZUIFDIBSUBSFJODMVEFE
0GDPVSTFIFLOFXIPXUPNBLFUPBTU)FECFFO 5IJTDIBSUEFNPOTUSBUFTUIBUXIBUFWFSJTCFJOHEPOF
XBUDIJOHIJTNPUIFSEPJUGPSZFBST#SFBEJTQMBDFE UPGJOEBOEGJYUIFSPPUDBVTFTPGEFWJBUJPOT JUJT
JOUIFUPBTUFS0ODFUIFCSFBEQPQQFEPVU UIFUPBTU GBJMJOH6OUJMBNPSFFGGFDUJWFBQQSPBDIJTBEPQUFE
XBTUBLFOUPUIFTJOLXIFSFUIFCVSOUBSFBTXFSF OPUIJOHNVDIXJMMJNQSPWF8PSLJOHIBSEFSBU
scraped off. existing approaches will fail (1).
5IFCPZDPVMEJNBHJOFOPPUIFSXBZUPNBLF
UPBTU5IJTBQQSPBDIXBTJOWBSJBCMZUIFPOFVTFECZ VARIATION IN QUALITY ATTRIBUTES
IJTNPUIFS)FDPVMEJNBHJOFOPPUIFSBQQSPBDI "TJNJMBSTJUVBUJPOFYJTUTGPSRVBMJUZDIBSBDUFSJTUJDT
JOEFWFMPQNFOUBOENBOVGBDUVSJOH7BSJBCMFTWBSZ
The Moral Of The Story 5IFKPCJTUPSFEVDFWBSJBUJPO*ONBOZQMBOUT
.BOZQFPQMFNJTTUIFNPSBMCFIJOEUIFPQFOJOH UIJTWBSJBUJPOJTJHOPSFEVOMFTTBTQFDJGJDBUJPOJT
TUPSZUPUIJTTFDUJPO-JLFUIFCPZ UIFZIBWF CSFBDIFE0DDBTJPOBMMZ BO004SFTVMUJTSFDPSEFE
EJGGJDVMUZJNBHJOJOHBOFOWJSPONFOUXIFSF (JWFOUIBUUIFSBUFPGEFWJBUJPOTJTSFBTPOBCMZ
EFWJBUJPOTSBSFMZIBQQFO5IFJSXPSMEJTPOFXIFSF TUBCMFBUNPTUTJUFT POFJTJODMJOFEUPDPODMVEFUIBU
UIFQSPDFTTSFNBJOTDPNQMJBOUCZSBJTJOHBOE UIFQSPDFTTFTJOWPMWFEBSFOPUCFJOHTVDDFTTGVMMZ
DMPTJOHPVUEFWJBUJPOT3FNBJOJOHDPNQMJBOUJT JNQSPWFE(FOVJOFJNQSPWFNFOUIFSFSFRVJSFT
JNQPSUBOUJOBSFHVMBUFEFOWJSPONFOU CVUJUJTOPU UIBUWBSJBUJPOJOUIFLFZWBSJBCMFTCFSFEVDFEOPU
OFDFTTBSZUPSBJTFBOEUIFODMPTFPVUEFWJBUJPOTUP NBOBHFE CVUSFEVDFE
SFNBJODPNQMJBOU$PNQMJBODFDPVMECFBDIJFWFE 0OFFYBNQMFDPNFTGSPNUIFGFSNFOUBUJPO
CZSFEVDJOHEFWJBUJPOTUPBOBCTPMVUFNJOJNVN QSPDFTTPGBESVHNBOVGBDUVSFS0WFSNBOZZFBST
5IFNPSBMPGUIFPQFOJOHTUPSZJTUIBUFWFSZXFFL BUUFNQUTIBECFFONBEFUPTPMWFQSPCMFNTJOUIF
UIFQIBSNBDFVUJDBMJOEVTUSZTQFOETNJMMJPOTPG GFSNFOUBUJPOQSPDFTT.JMMJPOTPGEPMMBSTIBE
EPMMBSTTDSBQJOHCVSOUUPBTU.VDIPGUIJTDPTUJT CFFOTQFOUPOTPQIJTUJDBUFEBVUPNBUJDDPOUSPM
wasteful and unnecessary. FRVJQNFOU4BEMZ OPSFBMQSPHSFTTXBTNBEF
&TTFOUJBMMZ UIFSFBSFUXPXBZTUPSFNBJO 5IFOBTUFBEZTUBUFUSJBMXBTDPOEVDUFE&WFSZ
DPNQMJBOU5IFGJSTUJTUPDBQUVSFUIFEFWJBUJPOT WBSJBCMFXBTLFQUBTDPOTUBOUBTQPTTJCMFGPS
BOEUPJOWFTUJHBUFUIFNUPFOTVSFUIBUOFJUIFSUIF BQQSPYJNBUFMZCBUDIFT*OTUFBEPGCFJOHDPOUFOU
QSPEVDUOPSUIFQBUJFOUXJMMCFDPNQSPNJTFE BOE UPSFNBJOXJUIJOBMMPXBCMFSBOHFT BOPQUJNVN
UPUBLFBQQSPQSJBUFDPSSFDUJWFBDUJPO5IFTFDPOEJT XBTBJNFEBUGPSFWFSZCBUDI#SVUBMMZVOJGPSN
UPSFEVDFWBSJBUJPOTPUIBUEFWJBUJPOTPGBOZLJOE PQFSBUJOHQSPDFEVSFTXFSFQPMJDFECZTFOJPS

Figure 1: Control chart for deviations.
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
PQFSBUPSTXIPXFSFHJWFOUIJTKPC5IFSFTVMUTXFSF PVUTJUVBUJPOTPSUIBUXIFOXFUSJFEUPJODSFBTF
JNNFEJBUFBOETQFDUBDVMBS BTOPUFEJO'JHVSF UISPVHIQVUWPMVNF BMMUIBUXFBDIJFWFEXBTBHMVU
*OFWFSZDSJUJDBMUPRVBMJUZDIBSBDUFSJTUJD WBSJBUJPO of work in progress.
ESPQQFECZBUMFBTU "CSJFGMPPLBUUIFGPSNVMBTIPXTXIZ5IFTF
8IBUUIFTUFBEZTUBUFUSJBMEFNPOTUSBUFEXBT DIBSBDUFSJTUJDTBSFJOUFSDPOOFDUFEJOFYPSBCMZ
that the central issues were not technical in their FONFTIFE5PVOEFSTUBOEBOEJNQSPWFPOF
OBUVSF UIFZXFSFPQFSBUJPOBM.PTUPGUIFXPSL PGUIFN JUXJMMCFOFDFTTBSZUPVOEFSTUBOEUIF
EPOFUPVOEFSTUBOEBOESFEVDFWBSJBUJPOXBTEPOF SFMBUJPOTIJQTUIBUFYJTUCFUXFFOBMMPGUIFN
CZGSPOUMJOFTVQFSWJTPSTBOEPQFSBUPST5IFZXFSF
HJWFOBTJOHMFBJN8JUIJOUIFMJNJUTPG(.1 Is Cycle Time Reduction The Answer? Or
UIFZXFSFUPNBLFFWFSZWBSJBCMFJOFWFSZCBUDIBT ERP Systems?
SFQFBUBCMFBOEQSFEJDUBCMFBTQPTTJCMF5IFFOFNZ 4PNFPGUIFMJUFSBUVSFDMBJNTUIBUUIFBOTXFSMJFT
XBTWBSJBUJPO BOEUIFJSKPCXBTUPDPORVFSJU5IFZ JOSFEVDJOHDZDMFUJNF"HBJO UIFGPSNVMBTIPXT
EJETP JOTQFDUBDVMBSGBTIJPO why. If work in progress is held constant and cycle
UJNFJTSFEVDFE UISPVHIQVUWPMVNFNVTUJODSFBTF
LITTLES LAWVARIATION IN VOLUME "MUFSOBUJWFMZ JGDZDMFUJNFJTSFEVDFECVUUIFSFJTOP
8FOFFEUPVOEFSTUBOEi-JUUMFT-BXw
UP OFFEUPJODSFBTFUISPVHIQVUWPMVNF XPSLJOQSPHSFTT
VOEFSTUBOEIPXWBSJBUJPOJNQBDUTPOQSPEVDUJPO BOEUIFSFGPSFDBSSZJOHDPTUT
DBOCFSFEVDFE
WPMVNF5IJTMBXDBOCFFYQSFTTFECZUIFGPMMPXJOH "OPUIFSXBZUPFYQSFTTDZDMFUJNFJTSFTJEFODFUJNF
GPSNVMB PSUIFMFOHUIPGUJNFUIFNBUFSJBMPSDPNQPOFOUT
SFTJEFJOUIFQSPDFTT$MFBSMZ JGUIJTJTSFEVDFE XPSLJT
"WFSBHF5ISPVHIQVU7PMVNF"WFSBHF8PSL*O GMPXJOHUISPVHIUIFQSPDFTTGBTUFS"MTP -JUUMFT-BX
1SPHSFTT"WFSBHF$ZDMF5JNF TFFNTUPJOEJDBUFUIBUJGXFDBOTJHOJGJDBOUMZSFEVDF
DZDMFUJNF XFHFUBMPUPGDPOUSPMPWFSUISPVHIQVU
#FBSJONJOEUIBU-JUUMFT-BXJTMBX OPUNFSFMZ WPMVNFBOEXPSLJOQSPHSFTT BOEUIJTJTDPSSFDU
BHPPEJEFB&WFSZGBDUPSZJTVOJRVF XJUIJUTPXO 4BEMZ BDPNNPOPVUDPNFGSPNUIJTSFWFMBUJPOJT
DPOTUSBJOUT MJNJUBUJPOT BOEDBQBCJMJUJFT/POFUIFMFTT that corporations put in place lowered targets for cycle
UIFSFMBUJPOTIJQFYQSFTTFEJO-JUUMFT-BXJTSFNBSLBCMZ UJNF*ONBOZDBTFT UIFQFPQMFJOUIFGBDUPSZIBWF
SPCVTU BOEJUBMXBZTBQQMJFT*UJTJNQPSUBOUUPOPUF GFXJEFBTPOIPXUPSFEVDFDZDMFUJNFXJUIPVUTJNQMZ
UIBUUIFUFSNTJO-JUUMFT-BXBSFBWFSBHFT8FDBO ESJWJOHQFPQMFBOEFRVJQNFOUIBSEFS8IFOUIJTJT
BMXBZTSFEVDFWBSJBUJPO CVUXFDBOOFWFSFMJNJOBUFJU EPOF TPNFDPNNPOPVUDPNFTBSFBSJTFJOEFWJBUJPOT
TPUIFUFSNTNVTUCFBWFSBHFT
EPXOUJNF BOEPUIFSQSPDFTTEJTUVSCBODFT XIJDIBSF
"NBKPSQSPCMFNJTUIBUNBOZQFPQMFDBOCFGPVOE MJLFMZUPJODSFBTFWBSJBUJPO DBVTFEFMJWFSZEFMBZT BOE
USZJOHUPJTPMBUFBOEJNQSPWFEFWJBUJPOT UISPVHIQVU EFHSBEFRVBMJUZ
WPMVNF XPSLJOQSPHSFTT PSDZDMFUJNFBTTFQBSBUF &OUFSQSJTFSFTPVSDFQMBOOJOH &31
TZTUFNTBSF
JTTVFT*OWFOUPSZJTBOPUIFSBTQFDUXJUIXIJDIQFPQMF XJEFTQSFBE4PNFDPNQBOJFTTXFBSCZUIFN BOE
TUSVHHMF BOEJUJTDMPTFMZBTTPDJBUFEXJUIUIFTFGBDFUT PUIFSTTXFBSBUUIFN)PXFWFS PODFXFOPUFUIBU
"OZBQQSPBDIUIBUBUUFNQUTUPJTPMBUFBOE UIFZBMMSFRVJSFTPNFTPSUPGQSFEJDUJPOTUPCFNBEF
JNQSPWFBOZPOFPGUIFTFDIBSBDUFSJTUJDTJTMJLFMZ BOEUIBUQSFEJDUBCJMJUZJTBGVODUJPOPGWBSJBUJPO JU
UPTUSVHHMF*GUIFCVTJOFTTJTBMSFBEZJOQPPS CFDPNFTDMFBSUIBUBOZ&31TZTUFNXJMMTUSVHHMFJO
TIBQF JUNJHIUCFQPTTJCMFUPNBLFTPNFJOJUJBM BIJHIMZWBSJBCMFFOWJSPONFOU
BEWBODFT#VUWFSZTPPO XFOPUFUIBUBUUFNQUTUP
reduce work in progress usually result in a lowering Littles Law Corollary
PGUISPVHIQVUWPMVNF MBUFEFMJWFSJFT PSTUPDL 5IFGJSTUDPSPMMBSZUP-JUUMFT-BXJTUIBUDZDMFUJNFJT

Figure 2: Reduced variation in potency (1).
400
350
300
250
200
150
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
BGVODUJPOPGWBSJBUJPOJOWPMVNF PSJGZPVQSFGFS necessary to create an approach that disentangles

WBSJBUJPOJOGMPXSBUFUISPVHIUIFQSPDFTT5IJTJTB UIFNJEEMFNBOBHFSTBOEUFDIOJDBMGPMLGSPN
GVOEBNFOUBMQSJODJQMFPGPQFSBUJPOTSFTFBSDI
UIFLOPUPGEBZUPEBZJTTVFTTPUIFZDBOGPDVT
"MMGBDUPSJFTIBWFQIZTJDBMDPOTUSBJOUTBOE POSFEVDJOHWBSJBUJPO"GUFSTUVEZJOHNBOZ
MJNJUBUJPOT'PSJOTUBODF UIFSFXJMMCFBMJNJUPO BQQSPBDIFT POMZPOFUIBUIBTFOKPZFEBQSFEJDUBCMZ
UIFBNPVOUPGXPSLJOQSPHSFTTUIBUDBOCFIFME IJHIMFWFMPGTVDDFTTXJMMCFEJTDVTTFE5IFSF
BUBOZHJWFOXPSLTUBUJPO4P JOBIJHIMZWBSJBCMF BSFPUIFST CVUTQBDFMJNJUBUJPOTQSFDMVEFBNPSF
QSPDFTT BCVJMEVQPGXPSLJOQSPHSFTTDBOIBMU DPNQSFIFOTJWFDPWFSBHF
QSPEVDUJPOJOVQTUSFBNTUFQTJOUIFQSPDFTT )PXFWFS MFUVTDPNNFODFXJUIBOBQQSPBDI
-JUUMFT-BXHJWFTVTBXBZPVU SFHBSEMFTTPG UIBU CZJUTFMG EPFTOPUXPSL4BEMZ JUJTBMTPWFSZ
XIFUIFSPVSQSPCMFNTBSFUISPVHIQVUWPMVNF DPNNPO*UJTBUSBJOJOHCBTFEBQQSPBDIXIFSF
XPSLJOQSPHSFTT PSDZDMFUJNF3FEVDFEWBSJBUJPO NBTTUSBJOJOHJTDPOEVDUFEBOEBQMFBJTNBEF
JOUIFGMPXPGNBUFSJBMTUISPVHIBQSPDFTTBMXBZT CZTFOJPSFYFDVUJWFTBUUIFFOEPGFBDIDPVSTFGPS
MPXFSTDZDMFUJNFGPSBOZHJWFOUISPVHIQVU QFSTPOOFMUPHPGPSUIBOEEPCBUUMFXJUIWBSJBUJPO
WPMVNF5IFSFGPSF SFEVDFEWBSJBUJPOXJMMJODSFBTF 5IFSFJTOPUIJOHJOUSJOTJDBMMZXSPOHXJUIMBSHF
UISPVHIQVUWPMVNFDBQBDJUZJGDZDMFUJNFGBMMT TDBMFUSBJOJOH*UDBOCFJOWBMVBCMFJODSFBUJOH
and work in progress is held steady. This in turn DPNNPOVOEFSTUBOEJOHT BTIBSFEFYQFSJFODF B
QSPWJEFTBOPQQPSUVOJUZUPSFEVDFVOJUDPTUT DPNNPOMBOHVBHF BOEJOQSPWJEJOHOFXTLJMMT#VU
*OQBSUJDVMBS WBSJBUJPOJOUIFJOQVUTUPBQSPDFTT CZJUTFMG JUTFMEPNCSJOHTBCPVUTJHOJGJDBOUDIBOHF
PSJOUIFGJSTUGFXFWFOUT HFOFSBMMZIBTUIFHSFBUFTU 4PNFUIJOHNPSFJTOFFEFE
JNQBDUPODZDMFUJNF XPSLJOQSPHSFTT BOE
UISPVHIQVUWPMVNF"OEZFUJUJTNPTUDPNNPOUP Steady State Trial
GJOEPVSNPTUUBMFOUFEQFPQMFDMVTUFSFEBSPVOEUIF "TUFBEZTUBUFUSJBMJTBQMBOUUSJBMUIBUIBTBTJUTBJN
USBOTGPSNBUJPOBMFWFOUJOUIFQSPDFTTPSUIBUQBSU UIFIPMEJOHPGFWFSZDPODFJWBCMFWBSJBCMFDPOTUBOU
of the process that does the core work the process GPSBOZHJWFOQSPEVDU3BXNBUFSJBMT NBDIJOF
XBTDSFBUFEUPEP6TVBMMZ UIFZBSFUPPMBUFJOUIF TFUVQT PQFSBUJOHQSPDFEVSFT UFNQFSBUVSFT
process. QSFTTVSFT GMPXSBUFT BOEBMMPUIFSWBSJBCMFTBSF
,OPXJOHUIBUSFEVDFEDZDMFUJNFXJMMQSPWJEFB LFQUBTDPOTUBOUBTQPTTJCMFGPSBOZHJWFOQSPEVDU
DPNQFUJUJWFBEWBOUBHFJTPOFUIJOH,OPXJOHXIBU *OTUFBEPGCFJOHDPOUFOUUPSFNBJOXJUIJOBMMPXBCMF
UPEPUPSFEVDFDZDMFUJNFXJUIPVUQVTIJOHUIF SBOHFT PQUJNVNWBMVFTGPSFBDIWBSJBCMFXFSF
QSPDFTTQBTUCSFBLJOHQPJOUJTBOPUIFS*OUIJTDBTF FTUBCMJTIFEBOEFWFSZBUUFNQUXBTNBEFUPIJUUIFTF
XFBSFUBMLJOHBCPVUSFEVDJOHWBSJBUJPOJOWPMVNF WBMVFT CBUDIBGUFSCBUDI
BTPQQPTFEUPSFEVDJOHWBSJBUJPOJOQSPEVDU 5IJTJTXIBUMFEUPUIFSFTVMUTFFOJO'JHVSF"MM
DIBSBDUFSJTUJDT PSRVBMJUZ
4P IPXEPXFEPUIBU NBOBHFSTBOEUFBNMFBEFSTXFSFMFGUJOOPEPVCU
UIBUDPORVFSJOHWBSJBUJPOUPPLQSFDFEFODFPWFSBMM
A PROVEN APPROACH FMTF5IJTJTPOFPGUIPTFUIJOHTUIBUJTFBTZUPTBZ
:PVNJHIUUIJOLUIBUSFEVDJOHWBSJBUJPOJT CVUNVDINPSFEJGGJDVMUUPEP*UTFWFSFMZUFTUTUIF
JNQFSBUJWF.BZCFZPVDBOBMTPDPOWJODFBGFX DPVSBHFPGXIPTPFWFSJTTVFTUIFJOTUSVDUJPO
FYFDVUJWFT CVUUIJTJTOPUFOPVHI4PNFIPXJUJT

#FDBVTFUIFQMBOUNBOBHFSOFWFSXBWFSFE 4PXFOPUFUIBUQPPSRVBMJUZBOESFMJBCJMJUZBSF
QFSTPOOFMTUBZFEGPDVTFEPOSFEVDJOHWBSJBUJPO DIJFGBNPOHTUUIFDBVTFTPGWBSJBUJPOJOWPMVNF
PODSFBUJOHSFQFBUBCJMJUZ5IFBQQSPBDIXPSLFE BOE UIFSFGPSF PGIJHIXPSLJOQSPHSFTT MPX
$ZDMFUJNFGFMM BTEJEXPSLJOQSPHSFTT5ISPVHIQVU UISPVHIQVUWPMVNFT BOEIJHIFSVOJUDPTUT-JUUMFT
WPMVNFDBQBDJUZSPTF:JFMETJODSFBTFEBOE -BXJTNPSFUIBOBHPPEJEFB*UJTBMBX*UJTBT
RVBMJUZJNQSPWFE(JWFOUIBUWBSJBUJPOJOFWFSZ VOGPSHJWJOHBOEVOJWFSTBMBTBSFUIFMBXTPGHSBWJUZ
DIBSBDUFSJTUJDXBTTJHOJGJDBOUMZMPXFSFE JUJTEJGGJDVMU BOEUIFSNPEZOBNJDT6OEFSTUBOEJOHUIJTMBXBOE
UPJNBHJOFBOZPUIFSPVUDPNF UIFFGGFDUPGWBSJBUJPODBOMFBEUPMBSHFTDBMFDPTU
0ODFMPXMFWFMTPGWBSJBUJPOBSFBDIJFWFE UIF SFEVDUJPOBOEJNQSFTTJWFJODSFBTFTJOQSPEVDUJPO
technical folk find that isolating and understanding
DBVTBMSFMBUJPOTIJQTJTNVDITJNQMJGJFE BOEGVSUIFS CONCLUSION
JNQSPWFNFOUTGMPX"LFZDIBSBDUFSJTUJDPGUIJT 5IFSFJTOPUSBEFPGGCFUXFFORVBMJUZ RVBOUJUZ
BQQSPBDIJTUIBUCFDBVTFJUJTBGPSNBMQMBOUUSJBM BOEDPNQMJBODF4PNFTJNQMFTUBUJTUJDBMDPODFQUT
DPNQMFUFXJUIEFBEMJOFT BTUSPOHGPDVTPOUIFKPCBU EFNPOTUSBUFUIBUPOMZPQFSBUJPOTUIBUTVDDFTTGVMMZ
IBOEJTBDIJFWFE SFEVDFWBSJBUJPOBOEJNQSPWFRVBMJUZXJMMBDIJFWF
IJHIPVUQVU MPXXPSLJOQSPHSFTT BOEHPPE
QUALITY AND QUANTITY DPNQMJBODF*NQSPWFERVBMJUZDSFBUFTSFEVDFE
$IJFGBNPOHTUDBVTFTPGWBSJBCJMJUZJOWPMVNFBSF EFWJBUJPOT TDSBQBOESFXPSL MPXFSXPSLJO
RVBMJUZBOESFMJBCJMJUZQSPCMFNT5IFMBUUFSBSF QSPHSFTTBOEJOWFOUPSJFT IJHIFSUISPVHIQVU
PCWJPVT*GCSFBLEPXOTBOEPUIFSEJTUVSCBODFT WPMVNF MPXFSVOJUDPTUT BOEFOIBODFEDVTUPNFS
PDDVS WBSJBCJMJUZJODZDMFUJNFJTJOFWJUBCMF TFSWJDFMFWFMT5IFHPPEOFXTJTUIBUXFDBOIBWFJU
1SPEVDURVBMJUZBMTPIFBWJMZJNQBDUTWBSJBCJMJUZ BMMJGPOMZXFLOPXIPXUPSFEVDFWBSJBCJMJUZ
JOWPMVNF*GQFPQMFBOENBDIJOFTNVTUTUSVHHMF
UPPCUBJOBTBUJTGBDUPSZSFTVMU WBSJBUJPOJOXPSL
SBUFT BOEUIFSFGPSF UISPVHIQVUWPMVNFJTBMNPTU REFERENCES
JOFWJUBCMF # /VOOBMMZ BOE + .D$POOFMM Six Sigma in the
004SFTVMUT SFXPSL TDSBQ BOEQPPSDIBOHFPWFST Pharmaceutical Industry $3$1SFTT
BOETFUVQIBWFBTJNJMBSFGGFDU2VJUFBQBSUGSPN 8)PQQ.4QFBSNBO Factory Physics .D(SBX)JMM
UIFPCWJPVTDPTUPGQPPSRVBMJUZ XFTVGGFSBMTPBO
JODSFBTFJOWBSJBUJPOJOWPMVNF5IJTBEEJUJPOBM
DPTUBMNPTUOFWFSBQQFBSTJODPTUPGRVBMJUZ ARTICLE ACRONYM LISTING
BOBMZTFT BOEZFUJUJTOFBSMZBMXBZTTJHOJGJDBOU ERP Enterprise Resource Planning
*OUFSFTUJOHMZ UIFGJSTUTVDDFTTGVMBQQMJDBUJPOTPG GMP Good Manufacturing Practice
UIJTBQQSPBDIUIFDPPSEJOBUPSPGUIJTDPMVNOTBX OOS Out of Specification
XBTJO.FUSPQPMJUBO1FSNBOFOU#VJMEJOH4PDJFUZ
CFGPSFJUCFDBNF.FUXBZ#BOL.S)PXBSE
.BOOJOHMFETFWFSBMQSPKFDUTUIBUTJHOJGJDBOUMZ
SFEVDFEDZDMFUJNFGPSBTQFDUTTVDIBTQSPEVDUJPOPG
NPOUIMZBDDPVOUTBOEMPBOBQQMJDBUJPOUVSOBSPVOE
UJNF*UXPSLTKVTUBTXFMMJOTFSWJDFBTJUEPFTJO
NBOVGBDUVSJOH

Paul L. Pluta and Timothy J. Fields
Substandard Data and

Documentation Practices
Case Study #5
Validation Case Studies discusses actual t.BKPSEBUBBOEEPDVNFOUBUJPOQSPCMFNTNBZ

validation situations useful to practitioners for CFDBUFHPSJ[FEBTEBUBSFDPSEJOHBOETUPSBHF
daily work applications. Each case presented deals EBUBSFTQPOTJCJMJUZBOEWFSJGJDBUJPO BOE
with a specific validation problem. A strategy and NJTDFMMBOFPVTEPDVNFOUBUJPOQSBDUJDFT
approach to solve the problem is proposed. The t3FDPNNFOEFEBDUJPOTUPBEESFTTUIFBCPWF
main objective of this column: Useful and practical QSPCMFNTBSFQSPQPTFE
information. t%FUBJMTJOWBMJEBUJPOQPMJDJFTBOEQSBDUJDFT
Reader comments, questions, and suggestions BEESFTTJOHEBUBBOEEPDVNFOUBUJPOQSPCMFNT
are needed to help us fulfill our objective for this are proposed.
column. Case studies illustrating validation issues t5SBJOJOHPOEBUBBOEEPDVNFOUBUJPOQSBDUJDFT
submitted by readers are most welcome. Please JTIJHIMZSFDPNNFOEFE5IJTUSBJOJOHTIPVMECF
send your comments and suggestions to journal periodically repeated.
coordinating editor Susan Haigney at shaigney@ t4FOJPSNBOBHFNFOUTVQQPSUPGBDUJWJUJFTUP
advanstar.com. BEESFTTEBUBBOEEPDVNFOUBUJPOQSPCMFNT
JTOFDFTTBSZGPSTVDDFTTGVMJNQSPWFNFOUPG
KEY POINTS DISCUSSED TVCTUBOEBSEQSBDUJDFT
The following key points are discussed: t7BMJEBUJPOBOEDPNQMJBODFQFSTPOOFMNVTU
tTest data and results generated as specified in DPOUJOVBMMZCFBXBSFPGUIFQPUFOUJBMGPS
UIFWBMJEBUJPOQSPUPDPMBSFDSJUJDBMCFDBVTFUIFZ EPDVNFOUBUJPOQSPCMFNTXIFOPSJHJOBMEBUBBSF
QSPWJEFUIFCBTJTGPSUIFGJOBMWBMJEBUJPOSFQPSU OPUSPVUJOFMZSFWJFXFE
BOEUIFKVEHNFOUUIBUUIFJUFNPGJOUFSFTUJT
WBMJEBUFE INTRODUCTION
t%BUBSFDPSEJOHBOEBTTPDJBUFEEPDVNFOUBUJPO "MMQSPDFTTFT FRVJQNFOU VUJMJUJFT BOBMZUJDBMNFUIPET
GPSWBMJEBUJPONBZOPUNFFUFYQFDUFERVBMJUZ FUD BTTPDJBUFEXJUINBOVGBDUVSJOHQIBSNBDFVUJDBM
standards. QSPEVDUTBOENFEJDBMEFWJDFTNVTUCFWBMJEBUFE
t4FWFSBMBDUVBMFYBNQMFTJOWPMWJOHQSPCMFNT *OHFOFSBM WBMJEBUJPODPNQSJTFTBTUBOEBSEJ[FE
XJUIEBUBBOEEPDVNFOUBUJPOBTTPDJBUFE EPDVNFOUBUJPOQSPDFTT5IFWBMJEBUJPONBTUFSQMBO
XJUIWBMJEBUJPOBSFEFTDSJCFE5IFTFJODMVEF JTBOPWFSWJFXPGUIFWBMJEBUJPOSFTQPOTJCJMJUJFTBUB
EBUBSFDPSEJOHPOTDSBQQBQFS FSSPOFPVT TJUF4QFDJGJDWBMJEBUJPOQMBOTJOJUJBUFBOJOEJWJEVBM
USBOTDSJCFEEBUB JNQSPQFSFSSPSDPSSFDUJPOT WBMJEBUJPO5IFQSPUPDPMTQFDJGJFTSFRVJSFEUBTLT
EBUBJODPOTJTUFOUXJUIQSPUPDPMT BOEPUIFS JODMVEJOHTBNQMJOHBOEUFTUJOH4BNQMJOHBOEUFTUJOH
FYBNQMFT SFTVMUTJOEPDVNFOUFEEBUBBOEPUIFSSFTVMUT5IF
[
ABOUT THE AUTHORS
For more Author Paul L. Pluta, Ph.D., has extensive pharmaceutical industry and university academic experience
information, and has been involved with the Journal of GXP Compliance and Journal of Validation Technology as
go to a writer and editor for several years. He may be reached at paul.pluta@comcast.net. Timothy J.
gxpandjvt.com/bios Fields is senior director of quality operations at Protein Sciences Corporation in Meriden, CT, with
responsibility for quality assurance and validation.

WBMJEBUJPOSFQPSUEJTDVTTFTSFTVMUTBOEQSPWJEFT Validation plan."GPSNBMSFRVFTUUPJOJUJBUF

DPODMVTJPOT5IFSFTQFDUJWFDPOUFOUPGUIFWBMJEBUJPO UIFWBMJEBUJPOQSPDFTTJTTVCNJUUFEUPUIFWBMJEBUJPO
QMBO QSPUPDPM SFTVMUT BOEWBMJEBUJPOSFQPSUNVTUCF BQQSPWBMDPNNJUUFF 7"$
%FWFMPQNFOUSFQPSUT
DPOTJTUFOUUPEFNPOTUSBUFUIBUUIFJUFNPGJOUFSFTUJT NBZCFJODMVEFEJOUIFSFRVFTUJOTVQQPSUPGUIF
WBMJEBUFE5FTUEBUBBOESFTVMUTHFOFSBUFEBTTQFDJGJFE DIBOHF5IFDIBOHFSFRVFTUJODMVEFTBQSPQPTFEMFWFM
JOUIFQSPUPDPMBSFDSJUJDBMJOUIFWBMJEBUJPOQSPDFTT PGXPSLUPDPOGJSNUIFBDDFQUBCJMJUZPGUIFDIBOHF
CFDBVTFUIFZQSPWJEFUIFUFDIOJDBMCBTJTGPSUIFGJOBM 5IFMFWFMPGXPSLJTCBTFEPOSJTLUPUIFQBUJFOUBOE
WBMJEBUJPOSFQPSU UPUIFPSHBOJ[BUJPO5IF7"$BQQSPWFTUIFDIBOHF
7BMJEBUJPONBOBHFSTGSPNTFWFSBMSFHVMBUFE SFRVFTU5IFBQQSPWFEDIBOHFSFRVFTUEPDVNFOUJT
DPNQBOJFTJOGPSNBMMZEJTDVTTFEQSPCMFNTBOE TUPSFEJOUIFWBMJEBUJPOMJCSBSZ
DPODFSOTBUUIF*OTUJUVUFPG7BMJEBUJPO5FDIOPMPHZT Protocol."QSPUPDPMJTXSJUUFOTQFDJGZJOH
Validation WeekDPOGFSFODFJO1IJMBEFMQIJB 1" JO EFUBJMFETBNQMJOHBOEUFTUJOHUPDPOGJSNUIF
0DUPCFS5IFNBOBHFSTFYQSFTTFEDPODFSO BDDFQUBCJMJUZPGUIFDIBOHF5IF7"$BQQSPWFTUIF
BCPVUUFTUJOHBOEEBUBHFOFSBUFEGPSUIFWBMJEBUJPO QSPUPDPM5IFBQQSPWFEQSPUPDPMJTTUPSFEJOUIF
5IFTFEBUBBOEEPDVNFOUTBSFQSPEVDFECZTDJFOUJTUT WBMJEBUJPOMJCSBSZ
FOHJOFFST BOEPUIFSUFDIOJDBMQFSTPOOFMXIPNBZ Validation work.7BMJEBUJPOXPSLJT
OPUCFJOWPMWFEXJUIHPPENBOVGBDUVSJOHQSBDUJDF QFSGPSNFEBDDPSEJOHUPUIFQSPUPDPM4BNQMJOHBOE
(.1
POBEBJMZCBTJT5IFJSEBUBSFDPSEJOHNBZOPU UFTUJOHBSFDPNQMFUFE%BUBBOEPUIFSSFTVMUTBSF
NFFUFYQFDUFERVBMJUZTUBOEBSET8JUIPVUQSPQFS generated and recorded.
EPDVNFOUFEEBUB BDDFQUBCMFWBMJEBUJPOSFQPSUT Validation report."SFQPSUDPOUBJOJOH
BSFOPUQPTTJCMF'VSUIFS TVCTUBOEBSEEBUBBOE all test results with discussion and conclusions is
EPDVNFOUBUJPOSFGMFDUQPPSMZPOUIFPSHBOJ[BUJPO QSFQBSFEBOETVCNJUUFEUPUIF7"$GPSBQQSPWBM5IF
UIJTJOUVSONBZIBWFTJHOJGJDBOUSBNJGJDBUJPOTJO SFQPSUJTBQQSPWFE BOEUIFQSPDFTTPSFRVJQNFOU
SFHVMBUPSZBVEJUT"DDFQUBCMFEBUBBOEBTTPDJBUFE DIBOHFJTJNQMFNFOUFE5IFBQQSPWFEWBMJEBUJPO
EPDVNFOUBUJPOGPSWBMJEBUJPOBSFDSJUJDBMMZJNQPSUBOU SFQPSUJTTUPSFEJOUIFWBMJEBUJPOMJCSBSZ
The discussion addresses situations in which Validation closure. If no other work is
PSJHJOBMEBUBBOEEPDVNFOUBUJPOHFOFSBUFEGPS OFFEFE UIFWBMJEBUJPOQSPKFDUJOJUJBUFECZUIFDIBOHF
WBMJEBUJPOSFQPSUTXFSFVOBDDFQUBCMF"WBSJFUZPG SFRVFTUJTDMPTFE
QSPCMFNTBSFEFTDSJCFE&YBNQMFTEFNPOTUSBUF Continued verification.1PTUWBMJEBUJPO
UIFCSPBETDPQFBOENBHOJUVEFPGQPUFOUJBM NPOJUPSJOHDPOGJSNJOHBDDFQUBCJMJUZPGUIFDIBOHF
QSPCMFNT&YBNQMFTEFTDSJCFEBSFBDUVBMPDDVSSFODFT continues throughout the lifecycle.
FYQFSJFODFECZUIFQBSUJDJQBUJOHNBOBHFST
This discussion will include the following: The issue addressed in this case study occurs
t#BDLHSPVOE5IFUZQJDBMWBMJEBUJPOPS JOTUFQOVNCFS5IFBDUVBMXPSLDPOEVDUFEUP
RVBMJGJDBUJPOQSPDFTTJTEFTDSJCFE DPOGJSNUIFBDDFQUBCJMJUZPGUIFWBMJEBUJPOQSPKFDUJT
t7BMJEBUJPOFWFOUT%FTDSJQUJPOTPGTFWFSBMBDUVBM EPDVNFOUFECZUFDIOJDBMQFPQMF%PDVNFOUBUJPO
WBMJEBUJPOFWFOUT NBZCFBDDPNQMJTIFEJOBWBSJFUZPGXBZTJODMVEJOH
t8IBUBSFUIFJTTVFT "TVNNBSZPGUIFQSPCMFNT SFTFBSDIOPUFCPPLT JOEJWJEVBMEBUBTIFFUT
EFTDSJCFE FMFDUSPOJDTZTUFNT BOEPUIFSBDDFQUBCMFNFUIPET
t$PSSFDUJWFBDUJPOTBOEQSFWFOUJWFBDUJPOT $"1"
5IFTFSFTVMUTBOEEBUBBSFBWBJMBCMFJOUIFWBMJEBUJPO
3FDPNNFOEFEBDUJPOTBOEJNQSPWFNFOUTUP SFQPSU FJUIFSBTDPQJFTPGPSJHJOBMEPDVNFOUBUJPO
QSFWFOUUIFTFQSPCMFNT PSSFUZQFEJOUPUIFSFQPSU0GUFO IPXFWFS
EPDVNFOUBUJPOJTTVCTUBOEBSEPSOPUBDDFQUBCMFGPSB
4QFDJGJDDPNNFOUTPOWBMJEBUJPOSFWJFXQSBDUJDFT WBSJFUZPGSFBTPOT
USBJOJOH BOENBOBHFNFOUTVQQPSUBSFBMTPQSPWJEFE
VALIDATION EVENTS
BACKGROUND "EJTDVTTJPOXJUITFWFSBMWBMJEBUJPONBOBHFST
5IFQSPDFTTPGWBMJEBUJPOUZQJDBMMZDPNQSJTFTUIF SFHBSEJOHQSPCMFNTXJUIWBMJEBUJPOBDUJWJUJFT
GPMMPXJOHTFRVFODFPGBDUJWJUJFT JOEJDBUFEUIBUUIFEPDVNFOUBUJPOPGPSJHJOBM
1. Change desired."OFRVJQNFOUPSQSPDFTT EBUB TUFQOVNCFS
XBTBTJHOJGJDBOUQSPCMFN
DIBOHFJTOFFEFEPSSFRVJSFE5IJTNBZCFB 4FWFSBMBDUVBMFYBNQMFTEFTDSJCFECZUIFNBOBHFST
OFDFTTBSZDIBOHFPSBEFTJSBCMFJNQSPWFNFOU EFNPOTUSBUFUIFNBHOJUVEFPGUIFQSPCMFN
Development work."QQSPQSJBUF 1FSTPOOFMSFDPSEJOHEBUBEJEOPUVOEFSTUBOEUIF
EFWFMPQNFOUXPSLJTDPNQMFUFEJOTVQQPSUPGUIF JNQPSUBODFPGUIFPSJHJOBMUFTUEBUBJOUIFWBMJEBUJPO
change. process.

Data Recording on Loose Paper 1SJNBSZEBUBTIPVMECFSFDPSEFEEJSFDUMZJOUP

7BMJEBUJPONBOBHFSTEFTDSJCFETJUVBUJPOTJOXIJDI UIFMBCPSBUPSZOPUFCPPLPSFMFDUSPOJDEBUBSFUSJFWBM
PSJHJOBMEBUBXFSFSFDPSEFEPOJOEJWJEVBMQJFDFTPG TZTUFNUIJTJTUIFQVSQPTFPGTVDITZTUFNT8IFO
iTDSBQQBQFSw JF VOCPVOE VOBUUBDIFE JOEJWJEVBM asked why data were not directly recorded into the
TIFFUTPGQBQFS
%BUBFJUIFSSFNBJOFEPOUIFTF MBCPSBUPSZOPUFCPPL IFSFTQPOEFEUIBUIFEJEOU
TIFFUTPSXBTDPQJFEJOUPMBCPSBUPSZOPUFCPPLT XBOUUPHFUUIFOPUFCPPLXFUPSEVTUZBTNJHIU
PSPUIFSEPDVNFOUBUJPOTZTUFNT*GUSBOTGFSSFEUP IBQQFOJOUIFNBOVGBDUVSJOHFOWJSPONFOUXIFSF
PUIFSTZTUFNT UIFPSJHJOBMEBUBTIFFUTXFSFVTVBMMZ the testing was done.
EJTDBSEFE1FPQMFSFDPSEJOHEBUBEJEOPUVOEFSTUBOE
UIBUXIFSFUIFZPSJHJOBMMZSFDPSEEBUBJTBQSJNBSZ Erroneous Transcribed Data
EPDVNFOUUIBUNVTUCFSFUBJOFE "GUFSDISPOPMPHJDBMMZPSHBOJ[JOHUIFiQBQFS
Data on paper bags.0OFWBMJEBUJPONBOBHFS UPXFMEBUBTIFFUTwEJTDVTTFEBCPWF PSJHJOBMEBUB
EFTDSJCFEBOFWFOUJOXIJDIBQBDLBHJOHQSPDFTT XFSFDPNQBSFEUPSFTVMUTMJTUFEJOUIFWBMJEBUJPO
FOHJOFFSTVCNJUUFEBOFYDFMMFOUSFQPSUEFTDSJCJOHUIF SFQPSU.PSFUIBOFSSPSTXFSFGPVOEOVNCFS
WBMJEBUJPODPOGPSNBODFSVOTGPSBCMJTUFSQBDLBHJOH USBOTQPTJUJPOT XSPOHOVNCFST BOEPUIFSQSPCMFNT
QSPDFTT5IFMJNJUTPGUIFQSPDFTTXFSFDPOGJSNFE $BSFMFTTMZXSJUUFO PSJHJOBMOVNCFSTDPOUSJCVUFE
1SPDFTTJOHBUUIFDFOUFSQPJOUXBTSFQMJDBUFEUP UPUIFQSPCMFN JF JTUIFOVNCFSBPSB
5IF
EFNPOTUSBUFSFQFBUBCJMJUZ5IFSFQPSUXBTXFMMXSJUUFO NBHOJUVEFPGUIFFSSPSTXBTOPUTVGGJDJFOUUPDIBOHF
BOEDPOUBJOFETFWFSBMSFMFWBOUJMMVTUSBUJPOT"MMEBUB UIFJOUFSQSFUBUJPOPGSFTVMUT4UJMM IPXFWFS UIF
XFSFTVQQMJFEJOBO&YDFMTQSFBETIFFUXJUITUBUJTUJDBM GJOBMWBMJEBUJPOSFQPSUEJEOPUBDDVSBUFMZSFGMFDUUIF
BOBMZTJTJOUIFGJOBMWBMJEBUJPOSFQPSU5IFSFTVMUT BDUVBMUFTUEBUB%BUBUSBOTGFSSFEUPUIFWBMJEBUJPO
QSPWFEUIFQSPDFTTXBTWBMJEBUFE5IF7"$XBTQMFBTFE report was not checked for accuracy. There were
XJUIUIFDPOUFOUBOERVBMJUZPGUIFWBMJEBUJPOSFQPSU OPTJHOBUVSFTCZQFSTPOOFMSFTQPOTJCMFGPSEBUB
5IFFOHJOFFSXBTUIFOBTLFEUPQSPWJEFUIFBDUVBM USBOTGFSPSWFSJGJDBUJPOPGUIFEBUBCZBTFDPOE
UFTUEBUBUIBUXFSFMJTUFEJOUIFSFQPSU)FSFUVSOFE QFSTPO$IFDLJOHEBUBUSBOTGFSJTUFEJPVTXPSL
UPUIFDPOGFSFODFSPPNXJUIBTFSJFTPGQBQFSCBHT OPUTPNFUIJOHUIBUDBOCFEPOFRVJDLMZPSXJUIB
DPOUBJOJOHCMJTUFSQBDLBHFTXJUIUFTUSFTVMUTXSJUUFOPO superficial approach.
UIFQBQFSCBHT5IFEBUBPOUIFCBHTIBECFFODPQJFE
JOUP&YDFM BQSJOUPVUPGXIJDIXBTBWBJMBCMFJOUIF Data Responsibility
WBMJEBUJPOSFQPSU 7BMJEBUJPONBOBHFSTSFQPSUFEUIBUQBHFTDPOUBJOJOH
5IFFOHJOFFSIBECFFOUSBJOFEPOQSPQFSSFDPSEJOH PSJHJOBMWBMJEBUJPOEBUBXFSFPGUFOOPUTJHOFEPS
PGEBUBJOBQFSNBOFOUCPVOEOPUFCPPL)PXFWFS EBUFECZQFSTPOOFMXIPSFDPSEFEUIFEBUB"MTP
UIFQSPEVDUJPOBSFBOFFEFEUIFNBDIJOFGPS UIFSFXBTOPWFSJGJDBUJPOPGEBUBCZBTFDPOEQFSTPO
DPNNFSDJBMQSPEVDUBOEUIFFOHJOFFSEJEOPUIBWF &YBNQMFTEFTDSJCFEBCPWF JF TDSBQQBQFS CBHT
UJNFUPQSFQBSFIJTOPUFCPPLUPQSPQFSMZSFDPSEUIF BOEQBQFSUPXFMFYBNQMFT
XFSFOPUTJHOFEPSEBUFE
UFTUSFTVMUT)FXSPUFUIFUFTUEBUBPOUIFCBHTEVFUP CZUIFFOHJOFFSOPSXFSFQBHFTWFSJGJFECZBTFDPOE
UJNFDPOTUSBJOUT QFSTPO8JUIPVUWFSJGJDBUJPOCZBTFDPOEQFSTPO
Data on paper towels."TJNJMBSFWFOUXBT there is no assurance that the work was actually
EFTDSJCFEJOWPMWJOHUBCMFUIBSEOFTTUFTUJOH"HBJOB done.
WFSZUIPSPVHISFQPSUDPOUBJOJOHIVOESFETPGUBCMFU
IBSEOFTTUFTUSFTVMUTXBTQSFTFOUFEGPSWBMJEBUJPO Error Corrections
BQQSPWBM#FDBVTFUIFSFXBTTPNVDIEBUBJOUIF 4FWFSBMWBMJEBUJPONBOBHFSTDPNNFOUFEPOFSSPS
SFQPSU UIF7"$BTLFEUPTFFUIFPSJHJOBMEBUBUPDIFDL DPSSFDUJPOTJOPSJHJOBMEBUB$IBOHFTBSFNBEF
GPSDPOTJTUFODZCFUXFFOUIFUZQFXSJUUFOSFQPSUBOEUIF CZiTDSBUDIPVUT wiDSPTTPVUT wVTJOHiXIJUFPVUw
original data. DPSSFDUJPOGMVJE BOEPUIFSNFBOT8IFONJTUBLFT
5IFGPSNVMBUJPOTDJFOUJTUUIFOQSFTFOUFEIJTXFMM BSFNBEF UIFZTIPVMECFDPSSFDUFECZMJOJOHPVU
PSHBOJ[FEBOEQFSGFDUMZDMFBOMBCPSBUPSZOPUFCPPL UIFFSSPOFPVTSFTVMUXJUIBTJOHMFMJOF BOEUIFO
UPUIF7"$8IFOBTLFEIPXIFXBTBCMFUPNBJOUBJO BEEJOHUIFDPSSFDUWBMVFOFBSUIFFSSPOFPVTWBMVF"
TVDIBOFBUOPUFCPPL UIFTDJFOUJTUEFTDSJCFEIJT DMFBSBOEDPODJTFFYQMBOBUJPOPGUIFNJTUBLFTIPVME
NFUIPEPGSFDPSEJOHEBUBPOTFQBSBUFTIFFUTPGQBQFS CFXSJUUFOOFBSUIFDIBOHFT7BMJEBUJPONBOBHFST
UIFOUSBOTGFSSJOHUIFEBUBUPIJTOPUFCPPL)FXBT DPNQMBJOFEUIBUFYQMBOBUJPOTPGNJTUBLFTXFSFPGUFO
UIFOBTLFEUPTIPXUIF7"$UIFPSJHJOBMEBUB)F DPNQMFUFMZBCTFOUPSXFSFQPPSMZXSJUUFO
SFUVSOFEXJUIBIBOEGVMPGXFUBOEDSVNQMFEQBQFS
UPXFMTUIBUIFXBTBCMFUPSFDPWFSGSPNUIFUSBTI5IFTF Data Incomplete
QBQFSUPXFMTDPOUBJOFEUIFUBCMFUIBSEOFTTUFTUdata. "WBMJEBUJPONBOBHFSEFTDSJCFEBOJODJEFOUJOXIJDI
BOFOHJOFFSXBTBTTJHOFEBQSPKFDUUPRVBMJGZGJWF

OFXNFUBMEFUFDUPSTUIBUSFQMBDFEPCTPMFUFNFUBM UFNQFSBUVSFPG$"DUVBMUFTUJOHJOEJDBUFEUIBUB
EFUFDUPSTBUUIFTJUF"WBMJEBUJPOQMBOXBTEFWFMPQFE NBYJNVNUFNQFSBUVSFPGPOMZ$XBTQPTTJCMF
BOEBQQSPWFECZUIFTJUF7"$'JWFJOEJWJEVBM 5IFFOHJOFFSDIBOHFEUIFQSPUPDPMGSPN$UP
JEFOUJDBMQSPUPDPMTXFSFQSFQBSFEBOEBQQSPWFE5IF $BOESFDPSEFEi1BTTwPOUIFEBUBQBHF$IBOHFT
GJSTUVOJUXBTUFTUFE BOEUIFWBMJEBUJPOSFQPSUXBT XFSFNBEFXJUIPVUDPOTVMUJOHTVQFSWJTPSTPS
XSJUUFO"MMEBUBXFSFUZQFXSJUUFOJOUPUIFSFQPSU JOWFTUJHBUJOHUIFSFBTPOGPSUIFQSPCMFN
5IF7"$BQQSPWFEUIFSFQPSU Missing equipment?"OPUIFSJODJEFOUXBT
5IFTFDPOEVOJUXBTUFTUFE BOEUIFWBMJEBUJPO EFTDSJCFEJOXIJDIUIFQSPUPDPMSFRVJSFEUFTUJOH
SFQPSUXBTXSJUUFO5IFWBMJEBUJPOQSPUPDPMGPSVOJU UISFFWJTJPOTZTUFNTJOTUBMMFEPOBQBDLBHJOH
OVNCFSUXPXBTJEFOUJDBMUPVOJUOVNCFSPOF CVU MJOF5IFFOHJOFFSDIBOHFEUIFSFRVJSFNFOUUP
BMMUFTUJOHTIPVMEIBWFCFFODPOEVDUFEVTJOHVOJU iUXPwTZTUFNT FYQMBJOJOHUIBUUIFMJOFPOMZIBE
OVNCFSUXP8IFOUIF7"$SFWJFXFEUIFEBUB UIFZ UXPTZTUFNTJOTUBMMFE BOEUIBUiUISFFwXBTB
OPUFEUIBUUIFUFTUEBUBTIFFUTXFSFMBCFMFEXJUI UZQPHSBQIJDBMFSSPS*OUSVUI UIFUIJSETZTUFNIBE
NFUBMEFUFDUPSOVNCFSPOFBUUIFUPQPGUIFQBHF CFFOTFOUUPUIFNBJOUFOBODFTIPQGPSSFQBJST5IF
5IFFOHJOFFSTBJEUIBUCFDBVTFUIFVOJUTXFSFBMMUIF FOHJOFFSXBTHPJOHPOWBDBUJPOBOEDPVMEOPUEFMBZ
TBNF UIFSFXBTOPOFFEUPSFQFBUUIFUFTUJOH5IF DPNQMFUJPOPGUIFWBMJEBUJPOXPSL5IFUIJSEWJTJPO
engineer knew that he was supposed to repeat the TZTUFNXBTFWFOUVBMMZRVBMJGJFE5IFGJOBMBQQSPWFE
UFTUJOHPOFBDIVOJU CVUXIFOIFUIPVHIUBCPVUJU EPDVNFOUUIFOPSJHJOBMMZDPOUBJOFESFRVJSFNFOUT
GPSBXIJMF IFEFDJEFEUIBUJUEJEOUNBLFTFOTFUP UPUFTUUISFFTZTUFNT DIBOHFTTUBUJOHUIBUPOMZUXP
EPTPNVDIUFTUJOH)FEFDJEFEUPVTFUIFEBUBGSPN WJTJPOTZTUFNTXFSFQSFTFOUPOUIFMJOF BOEGJOBMMZ
VOJUOVNCFSPOFJOUIFWBMJEBUJPOPGBMMUIFPUIFS changes to the change stating that the line contained
units. UISFFWJTJPOTZTUFNT5IJTEPDVNFOUXJUITPNBOZ
DIBOHFTBOEDPOGMJDUJOHFYQMBOBUJPOTXPVMEIBWF
Original Data Destroyed CFFOBOBVEJUiOJHIUNBSFwVOEPVCUFEMZSFGMFDUJOH
5IFJODPNQMFUFUFTUJOHQSPNQUFEUIF7"$UPBTL WFSZQPPSMZPOUIFFOHJOFFSBOEUIFPSHBOJ[BUJPO
UPTFFUIFPSJHJOBMUFTUEBUBGPSNFUBMEFUFDUPSVOJU
OVNCFSPOF5IFFOHJOFFSSFQMJFEUIBUCFDBVTFUIF Work Not Properly Stored and Not
WBMJEBUJPOXBTBQQSPWFE IFIBEEFTUSPZFEBMMIJT Retrievable
PSJHJOBMEBUB)FOFWFSLFQUPSJHJOBMEBUB"TMPOH "OPUIFSWBMJEBUJPONBOBHFSEFTDSJCFEBOJODJEFOU
BTUIFEBUBXFSFUZQFEJOUPUIFWBMJEBUJPOSFQPSU XIFSFCZBOBVEJUPSSFRVFTUFEBWBMJEBUJPOSFQPSU
UIFSFQPSUUIFOTFSWFEBTPSJHJOBMEBUB*GUIF7"$ GPSBOJNQBDUNJMM5IFWBMJEBUJPOSFQPSUXBT
XBOUFEUPTFFUIFPSJHJOBMEBUB UIFZTIPVMEIBWF BQQSPYJNBUFMZGJWFZFBSTPME5IFSFQPSUDPOUBJOFE
BTLFEIJNXIFOUIFZSFWJFXFENFUBMEFUFDUPS UZQFXSJUUFOEBUBPONBDIJOFSQNBOENBUFSJBM
OVNCFSPOF"OEJGUIF7"$IBEXBOUFEIJNUP TJFWFBOBMZTJTEFNPOTUSBUJOHNBDIJOFQFSGPSNBODF
SFUBJOUIFPSJHJOBMEBUB UIFZTIPVMEIBWFUPMEIJN The auditor asked to see the original test data. The
to do so. engineer who authored the report was asked to
SFUSJFWFIJTPSJHJOBMUFTUEBUB)FXBTOPUBCMFUPGJOE
Data Inconsistent with Protocol UIFPSJHJOBMEBUBTIFFUTBNPOHIJTQFSTPOBMGJMFT
7BMJEBUJPONBOBHFSTEFTDSJCFETJUVBUJPOTJOXIJDI /FJUIFSUIFFOHJOFFSOPSIJTEFQBSUNFOUIBEBEBUB
the original data did not agree with testing specified SFUSJFWBMTZTUFN)FTUPSFEIJTEBUBJOIJTQFSTPOBM
JOUIFBQQSPWFEQSPUPDPMTPNFSFBTPOBCMFBOE QSPKFDUGJMFTCVUIBENPWFEUPOFXPGGJDFMPDBUJPOT
PUIFSTDPNQMFUFMZVOBDDFQUBCMF TFWFSBMUJNFTEVSJOHUIFQSFWJPVTGJWFZFBST)F
Better testing.4FWFSBMNBOBHFSTEFTDSJCFE could not find the original data.
incidents in which the engineer executing the
protocol decided to change the testing specified Signature and Date Accuracy
JOUIFQSPUPDPM5IFSFXFSFNVMUJQMFSFBTPOTGPS "OPUIFSWBMJEBUJPONBOBHFSEFTDSJCFEBTJUVBUJPO
UIFDIBOHFT'PSFYBNQMF UIFUFTUJOHDPVMECF JOXIJDIBWBMJEBUJPOSFQPSUXBTCFJOHSFWJFXFECZ
EPOFNPSFFGGJDJFOUMZPSUIFUFTUJOHXPVMECFNPSF UIF7"$4PNFCMBOLTQBDFTXFSFPCTFSWFEPOEBUB
SFMFWBOUUPGVUVSFBDUVBMQFSGPSNBODF5IFTFDIBOHFT sheets containing original analytical data indicating
BSFCFTUIBOEMFECZBNFOEJOHUIFQSPUPDPMBOE NJTTJOHEBUB"OBMZUJDBMEBUBDPVMEOPUCFGPVOE
PCUBJOJOH7"$BQQSPWBMCFGPSFFYFDVUJPOPGUFTUT UIFMBCIBEOPUSFDFJWFEUIFTBNQMFT/FXTBNQMFT
Test failures..BOBHFSTBMTPEFTDSJCFEJODJEFOUT XFSFXJUIESBXOGSPNUIFMPUBOEXFSFUFTUFE5IF
in which the engineer changed the test procedure new test data were then added to the data sheets in
CFDBVTFUIFFRVJQNFOUGBJMFEUIFUFTUBTTQFDJGJFE UIFWBMJEBUJPOSFQPSU5IJTFOUJSFFWFOUDPOTVNFE
JOUIFQSPUPDPM'PSFYBNQMF UIFRVBMJGJDBUJPO BQQSPYJNBUFMZUXPXFFLTGPSSFTBNQMJOH UFTUJOH BOE
QSPUPDPMGPSBGMVJEJ[FECFEESZFSSFRVJSFEBOPVUMFU SFTVMUT)PXFWFS UIFEBUBTIFFUTXFSFTJHOFEBOE

EBUFEBTJGBMMUFTUJOHXBTDPNQMFUFEBUUIFUJNFPGUIF BSFEFTJSBCMFUIFZQSPWJEFTFDVSJUZBOEFOBCMF
PSJHJOBMTBNQMFT8IFOBTLFEBCPVUUIFJOBDDVSBUF JNNFEJBUFBDDFTT3BQJESFUSJFWBMPGEBUBJT
EBUFPOUIFSFQPSU UIFBOBMZTUSFTQPOEFEUIBUBMM FTQFDJBMMZJNQPSUBOUXIFOSFRVFTUFECZSFHVMBUPSZ
TBNQMFTTIPVMEIBWFCFFOUFTUFEPOUIBUEBUF TPUIBU auditors.
XBTUIFQSPQFSEBUFUPCFSFDPSEFE %BUBNBZCFSFDPSEFEJOUPCPVOEMBCPSBUPSZ
OPUFCPPLTXJUIDPOTFDVUJWFMZOVNCFSFEQBHFT
Electronic Data Transfer and Data Changes /VNCFSFEQBHFTBSFSFRVJSFEUPQSFWFOUQBHFTGSPN
"OPUIFSWBMJEBUJPONBOBHFSEFTDSJCFEBTJUVBUJPOJO CFJOHSFNPWFEGSPNUIFOPUFCPPLXJUIPVUCFJOH
XIJDIEBUBJOUIFWBMJEBUJPOSFQPSUXFSFNPEJGJFE OPUJDFE5ISFFSJOHCJOEFSTXJUIMPPTFQBHFTBSF
CZUSBOTGFSCFUXFFOXPSLTUBUJPOTBOEBTTPDJBUFE OPUTVJUBCMFGPSVTFBTMBCPSBUPSZOPUFCPPLT'JMMFE
DPNQVUFSQSPHSBNT4PGUXBSFNBZPWFSSPVOEEBUF OPUFCPPLTBOEOPUFCPPLTCFMPOHJOHUPQFSTPOOFM
USVODBUFEBUF PSNBLFPUIFSDIBOHFTBTQBSUPG XIPIBWFMFGUUIFPSHBOJ[BUJPOTIPVMECFTUPSFEJO
DBMDVMBUJPOT5IFTFDIBOHFTIBQQFOVOCFLOPXOTU BTFDVSFEBSFB*GOPUFCPPLTBSFSFNPWFEGSPNUIF
UPUIPTFVTJOHUIFQSPHSBNT5IFZNBZBEE TFDVSFEBSFB UIFSFNVTUCFBNFDIBOJTNUPSFUVSO
WBSJBUJPOUPUIFEBUBPSCJBTUIFEBUB UIFOPUFCPPLUPUIFBSFB JF VOSFTUSJDUFEBDDFTTBOE
5IFQSPCMFNEFTDSJCFEBCPWFJTPOFPGTPGUXBSF SFNPWBMPGJOGPSNBUJPOJTOPUBDDFQUBCMF
3FTFBSDI
WBMJEBUJPOBOECFZPOEUIFTDPQFPGUIJTBSUJDMF*U BOEEFWFMPQNFOUTDJFOUJTUTBSFBDDVTUPNFEUPVTJOH
JT IPXFWFS BOJNQPSUBOUDPODFSOXIFOEBUBBSF MBCPSBUPSZOPUFCPPLT)PXFWFS NBOVGBDUVSJOHTJUFT
USBOTGFSSFECFUXFFOJOEJWJEVBMTBOEDPNQVUFSTJO NBZOPUCFGBNJMJBSXJUIUIFTFQSBDUJDFTBOEOPU
UIFPSHBOJ[BUJPO XIFODPNQVUFSDBMDVMBUJPOTBSF IBWFTFDVSFTZTUFNTJOQMBDF&WFOXPSTF UIFZNBZ
QFSGPSNFECZNVMUJQMFJOEJWJEVBMT XIFODPNQVUFS CFSFTJTUBOUUPJNQMFNFOUJOHFWFOTJNQMFOPUFCPPL
DBMDVMBUJPOTNBZCFEPOFPOIPNFDPNQVUFSTZTUFNT TZTUFNT(SPVQTNBZCFMJFWFUIBUEPDVNFOUBUJPO
UIBUXFSFOPURVBMJGJFE BOEPUIFSQPTTJCMFTDFOBSJPT QSPDFEVSFTBSFOPUJNQPSUBOUPSBSFUPPSFTUSJDUJWFUP
UIFJSSFHVMBSSFTQPOTJCJMJUJFT
WHAT ARE THE PROBLEMS? 4ZTUFNTJOXIJDIUFDIOJDBMQFSTPOOFMBSFBTTJHOFE
5IFEBUBBOEEPDVNFOUBUJPOQSPCMFNTEFTDSJCFE TQFDJGJDOVNCFSFEJOEJWJEVBMOPUFCPPLTDBOCF
IFSFJONBZCFHFOFSBMMZHSPVQFEJOUPUIFGPMMPXJOH FBTJMZJNQMFNFOUFE"OPUFCPPLBENJOJTUSBUPS
GVOEBNFOUBMQSPCMFNT BUUIFTJUFTIPVMECFEFTJHOBUFE5IFOPUFCPPL
tData recording and storage."MMEBUBNVTUCF BENJOJTUSBUPSJTSFTQPOTJCMFGPSUSBDLJOHBMM
SFDPSEFEJOUPBTZTUFNUIBUJTDPNQMFUFMZTFDVSF OPUFCPPLTBUUIFTJUF/PUFCPPLTBTTJHOFEUP
BOEGSPNXIJDIUIFPSJHJOBMEBUBDBOCFSFUSJFWFE UFDIOJDBMQFSTPOOFMBSFVTFEBOETUPSFECZUIFN
XJUIJOBSFBTPOBCMFUJNFQFSJPE JOUIFJSPGGJDFTPSMBCPSBUPSJFT5IFZTIPVMECF
tData responsibility and verification." stored in locked desks or other secure locations.
SFTQPOTJCMFQFSTPOTIPVMETJHOBOEEBUFXIFOFWFS 8IFOJOEJWJEVBMTXIPIBWFOPUFCPPLTMFBWFUIF
EBUBBSFSFDPSEFEPSUSBOTGFSSFE"TFDPOEQFSTPO PSHBOJ[BUJPO UIFZNVTUSFUVSOUIFOPUFCPPLTUPUIF
TIPVMEBQQSPQSJBUFMZWFSJGZUIFBDUJPO TJUFOPUFCPPLBENJOJTUSBUPS5IJTFOBCMFTBMMEBUBJO
tMiscellaneous documentation practices. UIFOPUFCPPLUPBMXBZTCFBDDFTTJCMFUPBVEJUPSTPS
0SHBOJ[BUJPOTNVTUIBWFQSPDFEVSFTEFGJOJOH for other needs.
TJHOBUVSFT EBUFT DPSSFDUJOHEBUB FYQMBJOJOH 0SJHJOBMEBUBDBOOPUCFSFDPSEFEPOTDSBQQBQFS
DPSSFDUJPOT BOEBTTPDJBUFEDPOTJEFSBUJPOT BOETUPSFEJOBQFSTPOBMEFTLESBXFSJOBDPNQMJBOU
PSHBOJ[BUJPO
CORRECTIVE ACTION AND PREVENTIVE
ACTION Data Responsibility and Verification
7BMJEBUJPONBOBHFSTEJTDVTTFEUIFQSPCMFNT 8IFOFWFSEBUBBSFSFDPSEFE UIFQFSTPOSFDPSEJOH
QSFTFOUFEJOUIJTBSUJDMFBOESFDPNNFOEFETPMVUJPOT UIFEBUBNVTUBGGJSNUIFBDDVSBDZPGUIFEBUBBOE
UIBUUIFZIBWFTVDDFTTGVMMZJNQMFNFOUFE BDDFQUSFTQPOTJCJMJUZGPSUIFEBUB5IJTJTEPOFCZ
TJHOBUVSFBOEEBUFPOUIFPSJHJOBMEBUBEPDVNFOU
Data Recording and Storage 1FPQMFNVTUCFBDDPVOUBCMFGPSUIFJSXPSL8IFO
"MMEBUBNVTUCFSFDPSEFEJOUPBTZTUFNUIBUJT UIFZBSFSFRVJSFEUPTJHOBOEEBUFEPDVNFOUTPS
TFDVSF SFUSJFWBCMF BOEQSPUFDUJWFPGEBUBJOUFHSJUZ EBUBBGGJSNJOHUIFQFSGPSNBODFPGUIFJSXPSL
5IFTZTUFNNVTUBMMPXEBUBUPCFSFUSJFWFEBT UIFBDDVSBDZPGUIFJSEBUB BOETPPO JF UIFJS
OFFEFEXJUIJOBSFBTPOBCMFUJNFQFSJPE%BUB QFSTPOBMJOUFHSJUZJTCFIJOEUIFEPDVNFOU

NBZCFSFDPSEFEJOUPCPVOEMBCPSBUPSZOPUFCPPLT QFSTPOOFMSFWJFXJOHUIFEPDVNFOUIBWFNVDI
PSJOUPFMFDUSPOJDTZTUFNT&MFDUSPOJDTZTUFNT HSFBUFSDPOGJEFODFJOUIFDPOUFOUPGUIFEPDVNFOU

%FQFOEJOHPOUIFUZQFBOEDJSDVNTUBODFTPGUIF WBMJEBUJPOEPDVNFOUTNVTUOPUCFDPNQMFUFMZ
TJUVBUJPO PSJHJOBMEBUBTIPVMEBMTPCFXJUOFTTFECZ dependent and trusting of the data typed into
BTFDPOEQFSTPO7FSJGJDBUJPOCZBTFDPOEJOEJWJEVBM WBMJEBUJPOSFQPSUTPSPUIFSEPDVNFOUTDPOUBJOJOH
NBZCFMJLFOFEUP(.1SFRVJSFNFOUTUIBUDSJUJDBM transferred data.
TUFQTJONBOVGBDUVSJOHNVTUCFWFSJGJFECZBTFDPOE 3FHBSEJOHUSBOTGFSSFEEBUB 7"$NFNCFSTTIPVME
QFSTPO7FSJGJDBUJPOCZBTFDPOEJOEJWJEVBMBQQMJFT BMTPJOTJTUPOSFTQPOTJCJMJUZ TJHOBUVSFBOEEBUF
PG
to original data recording as well as to data transfer QFSTPOOFMXIPIBWFUSBOTDSJCFEEBUBGSPNPSJHJOBM
GSPNPSJHJOBMEBUBSFDPSETUPDPNQVUFSTZTUFNTPS EPDVNFOUBUJPOTUPUIFWBMJEBUJPOSFQPSU1PMJDJFT
UPWBMJEBUJPOSFQPSUT*G GPSFYBNQMF PSJHJOBMEBUB TIPVMETQFDJGZSFRVJSFNFOUTGPSEBUBWFSJGJDBUJPO"MM
BSFFOUFSFEJOUPBTQSFBETIFFUGPSTUBUJTUJDBMBOBMZTJT EBUBOVNCFSTTIPVMECFWFSJGJFEGPSTNBMMBNPVOUT
PSJHJOBMEBUBFOUSJFTPOUIFQSJOUFETQSFBETIFFUNVTU PGEBUB3FBTPOBCMFTBNQMJOHPGEBUBGPSWFSJGJDBUJPO
CFWFSJGJFEGPSBDDVSBDZBOEBGGJSNFECZTJHOBUVSF JTBDDFQUBCMFGPSMBSHFBNPVOUTPGEBUB*OBOZDBTF
and date on the printout of the spreadsheet. BDDFQUBODFPGUSBOTGFSSFEEBUBJOBWBMJEBUJPOSFQPSU
XJUIPVUWFSJGJDBUJPONVTUOPUPDDVSJOUIFBQQSPWBM
Miscellaneous Data and Documentation QSPDFTTGPSWBMJEBUJPOEPDVNFOUT
Practices
.JTDFMMBOFPVTEPDVNFOUBUJPOQSBDUJDFTNVTUCF Storage of Original Data in Validation
EFGJOFEJOQSPDFEVSFT5IJTBQQMJFTUPTJHOBUVSFT Documents
JOJUJBMT EBUFT DPSSFDUJOHFSSPST FYQMBJOJOHFSSPST 0SJHJOBMEBUBNBZCFTUPSFEBTQBSUPGUIFBQQSPWFE
EBUJOHQSBDUJDFT BOEPUIFSQSPDFEVSFTBTTPDJBUFE WBMJEBUJPOSFQPSUJOUIFWBMJEBUJPOMJCSBSZ5IF
XJUIEBUBSFDPSEJOHBOEWFSJGJDBUJPO5IFTFQSBDUJDFT WBMJEBUJPOMJCSBSZTIPVMECFBTFDVSFBSFBGSPN
BSFVTVBMMZXFMMEFGJOFEJONBOVGBDUVSJOHBOE XIJDIWBMJEBUJPOEPDVNFOUTDBOOPUCFSFNPWFE
RVBMJUZPSHBOJ[BUJPOT3FMBUFEPSHBOJ[BUJPOT *GBTJUFEPFTOPUIBWFBEPDVNFOUTUPSBHFTZTUFN
JF UIPTFOPUEJSFDUMZJOWPMWFEJOEBJMZ(.1 PSBCPVOEOPUFCPPLTZTUFNBOEJTVOXJMMJOHUP
BDUJWJUJFT
NBZOPUIBWFSJHPSPVTEPDVNFOUBUJPO JNQMFNFOUTVDIBTZTUFN TUPSBHFPGPSJHJOBMEBUB
QSPDFEVSFT.PSFJNQPSUBOUMZ NBOBHFNFOUPGUIFTF JOWBMJEBUJPOEPDVNFOUTTIPVMECFEPOF5IF
PSHBOJ[BUJPOTBOEUIFDVMUVSFPGUIFTFBSFBTNBZOPU WBMJEBUJPOSFQPSUNVTUCFTVQQPSUFECZPSJHJOBM
FNQIBTJ[FUIFJNQPSUBODFPGHPPEEPDVNFOUBUJPO EBUB*GUIFTJUFEPFTOPUIBWFBHPPEEBUBTUPSBHF
QSBDUJDFT0SHBOJ[BUJPOTFWFOSFNPUFMZDPOOFDUFE BOESFUSJFWBMTZTUFN UIFWBMJEBUJPOBSFBTIPVME
UP(.1BDUJWJUJFTNVTUIBWFQSPDFEVSFTEFGJOJOH TUPSFUIFPSJHJOBMEBUBXJUIJOUIFWBMJEBUJPO
UIFTFQSBDUJDFT1FSTPOOFMNVTUVOEFSTUBOEBOE package.
CFDPNQMJBOUXJUIQSPDFEVSFT.BOBHFNFOUNVTU
TVQQPSUDPNQMJBODFUPUIFTFQSPDFEVSFTJOQSBDUJDF TRAINING
and in the spoken word. 5IFSFDPNNFOEBUJPOTQSFTFOUFEJOUIJTBSUJDMFNBZ
OFDFTTJUBUFUIFEFWFMPQNFOUPSTUSFOHUIFOJOHPG
VALIDATION POLICY AND PRACTICE TFWFSBMQSPDFEVSFTJOUIFPSHBOJ[BUJPO5SBJOJOH
7BMJEBUJPONBOBHFSTTVHHFTUFEEFWFMPQNFOUPS POUIFTFQSPDFEVSFTJTNBOEBUPSZ5IFFYBNQMFT
FOIBODFNFOUPGEPDVNFOUBUJPOQSBDUJDFTBTTPDJBUFE QSPWJEFECZWBMJEBUJPONBOBHFSTJOEJDBUFE
XJUIWBMJEBUJPOEPDVNFOUT5IFGPMMPXJOHUXP TJHOJGJDBOUNJTVOEFSTUBOEJOHPGGVOEBNFOUBM
QSBDUJDFTXFSFSFDPNNFOEFEGPSWBMJEBUJPOHSPVQT QSPDFEVSFTBOETFSJPVTOPODPNQMJBODF
.BOVGBDUVSJOHBOEFOHJOFFSJOHQFSTPOOFMBSFPGUFO
Review of Original Data by the VAC as Part OPUBTGBNJMJBSXJUIEBUBSFDPSEJOHQSPDFEVSFT
of Validation Report Approval Processes BTRVBMJUZBTTVSBODF 2"
PS(.1BOBMZUJDBMMBC
8IFOUIF7"$JTSFWJFXJOHBWBMJEBUJPOSFQPSU QFSTPOOFM1SPDFEVSFTBEESFTTJOHEBUBSFDPSEJOH
DPOUBJOJOHUFTUEBUBPSPUIFSSFTVMUT UIF TJHOBUVSFT EBUFT BOETPPONBZCFJOGSFRVFOUMZ
EPDVNFOUBUJPODPOUBJOJOHPSJHJOBMEBUBNVTU VTFEJOUIFJSEBJMZKPCGVODUJPOT5SBJOJOHBOE
BMTPCFSFWJFXFE5IJTQSBDUJDFTIPVMECFDMFBSMZ QFSJPEJDSFUSBJOJOHNVTUCFDPOTJEFSFEEFQFOEJOH
TUBUFEJOBWBMJEBUJPOQSPDFEVSF*NQMFNFOUJOHUIJT POUIFPWFSBMMDPNQFUFODFPGUIFPSHBOJ[BUJPO
QSBDUJDFXJMMFWFOUVBMMZFMJNJOBUFEBUBSFDPSEJOH *OEJWJEVBMTXJUIHPPEEPDVNFOUBUJPOTLJMMTNBZ
POQBQFSUPXFMTBOETJNJMBSVOBDDFQUBCMFNFEJB*G CFBTTJHOFEUPXPSLXJUIQFSTPOOFMXIPBSFVOBCMF
PSJHJOBMEBUBBSFTUJMMSFDPSEFEPOJOEJWJEVBMTIFFUT UPNFFUUSBJOJOHTUBOEBSET JF BQBSUOFSTIJQ
PGQBQFS QBQFSUPXFMT PSQBQFSCBHT UIFTFNVTU BQQSPBDIUPGBDJMJUBUFBDDFQUBCMFEBUBSFDPSEJOH

CFDPOTJEFSFEQSJNBSZEPDVNFOUTBOECFSFUBJOFE 4PNFPSHBOJ[BUJPOTIBWFIJSFE(.1TQFDJBMJTUTGPS
BOETUPSFEBQQSPQSJBUFMZ5IFZNVTUCFSFWJFXFE this purpose.
JOBEEJUJPOUPUIFEBUBQSPWJEFEJOUIFWBMJEBUJPO 5IFTJUF7"$NVTUCFFTQFDJBMMZXFMMUSBJOFEJO
SFQPSU7"$NFNCFSTXIPSFWJFXBOEBQQSPWF UIFTFQSPDFEVSFT5IF7"$GVODUJPOTBTTVSSPHBUF

BVEJUPSTPGWBMJEBUJPOEPDVNFOUT5IFZNVTUOPU 7BMJEBUJPOBOEDPNQMJBODFQFSTPOOFMNVTUCF
BTTVNFUIBUBXFMMXSJUUFOWBMJEBUJPOSFQPSUJTCBTFE BXBSFPGUIFQPUFOUJBMGPSEPDVNFOUBUJPOQSPCMFNT
POBDDFQUBCMFPSJHJOBMEBUB3FWJFXPGPSJHJOBMEBUB when original data are reproduced in reports.
NVTUCFQBSUPGBOZSFQPSUSFWJFX "QQSPWBMPGEPDVNFOUTDPOUBJOJOHUSBOTDSJCFE
PSDPQJFEEBUBTIPVMEBMXBZTJODMVEFSFWJFXPG
SENIOR MANAGEMENT SUPPORT original data.
+VTUBTXJUIBOZNBKPSQPMJDZPSQSPDFEVSBM 7BMJEBUJPOBUBSFHVMBUFENBOVGBDUVSJOHTJUF
DIBOHFJOBOPSHBOJ[BUJPO TFOJPSNBOBHFNFOU JTBEJGGJDVMUKPC4VDDFTTGVMWBMJEBUJPOEFQFOET
NVTUCFTVQQPSUJWFPGDIBOHFTJOEBUBBOE POKPCQFSGPSNBODFBDDPSEJOHUPQSPDFEVSFTCZ
EPDVNFOUBUJPOQSBDUJDFT4FOJPSNBOBHFNFOU JOEJWJEVBMTXJUIEJGGFSFOUCBDLHSPVOET FYQFSUJTF
NBZOPUCFGBNJMJBSPSIBWFJOUFSFTUJOEBUBPS TLJMMT FYQFSJFODF BOEUSBJOJOH7BMJEBUJPOJTBMTP
EPDVNFOUBUJPOQSPCMFNT4FOJPSNBOBHFNFOU TVCKFDUUPUIFTUSFTTBOEQSFTTVSFPGUIFCVTJOFTT
NBZGPDVTPOGJOBODJBMJTTVFTPSPUIFSOPO(.1 FOWJSPONFOU'BNJMJBSJUZXJUIBMMSFMFWBOU
JTTVFT7BMJEBUJPOBOEDPNQMJBODFQFSTPOOFMNBZ QSPDFEVSFTBUBMMUJNFTDBOOPUCFBTTVNFE5IFTF
OFFEUPQFSTJTUJOFEVDBUJOHTFOJPSNBOBHFNFOU QSPCMFNTBSFFYBDFSCBUFEJOUJNFTPGSFEVDFE
BCPVUEBUBBOEEPDVNFOUBUJPOQSPCMFNT BOE IFBEDPVOU OFXQFSTPOOFM DPOUSBDUMBCPS UJNF
BCPVUUIFQPUFOUJBMSBNJGJDBUJPOTPGUIFTFQSPCMFNT DPOTUSBJOUT GJOBODJBMDPOTUSBJOUT CBDLPSEFST
UPUIFFOUJSFPSHBOJ[BUJPO&GGPSUTUPVQHSBEFEBUB QSPKFDUEFBEMJOFT JODFOUJWFHPBMT BOEBTTPDJBUFE
BOEEPDVNFOUBUJPOQSBDUJDFTJOUIFPSHBOJ[BUJPO occurrences.
UIBUSFRVJSFTJHOJGJDBOUDIBOHFTXJUIPVUTFOJPS %BUBBOEEPDVNFOUBUJPOQSPCMFNTBSFPOHPJOH
NBOBHFNFOUTVQQPSUXJMMCFGVUJMF QSPCMFNTUIFZOFWFSHPBXBZ7BMJEBUJPOBOE
DPNQMJBODFQFSTPOOFMTIPVMEQFSTJTUJOFODPVSBHJOH
CONCLUSIONS FMFDUSPOJDTZTUFNT PQUJNJ[FEEFTJHOPGGPSNT BOE
5IFDPOUFOUPGUIJTEJTDVTTJPOEFTDSJCFEB NJOJNJ[JOHPQQPSUVOJUJFTGPSQFSTPOOFMFSSPSTJO
NVMUJUVEFPGEPDVNFOUBUJPOQSPCMFNTBTTPDJBUFE EPDVNFOUBUJPO5SBJOJOHJTBOFDFTTJUZ5SBJOJOH
XJUIPSJHJOBMWBMJEBUJPOUFTUEBUB7BMJEBUJPO NVTUBMTPCFQFSJPEJDBMMZSFQFBUFEUPFNQIBTJ[F
EBUBBSFPGUFOHFOFSBUFECZUFDIOJDBMQFSTPOT and encourage proper practice.
PSFOHJOFFSTXIPBSFOPUGBNJMJBSXJUIUZQJDBM 7BMJEBUJPOBOEDPNQMJBODFQFSTPOOFMNVTU
(.1EPDVNFOUBUJPOQSBDUJDFT#FDBVTFUIFZBSF CFDPOUJOVBMMZWJHJMBOUJOBTTVSJOHUIBUPSJHJOBM
OPUJOWPMWFEXJUIDPNNFSDJBMNBOVGBDUVSJOHPS SFDPSEFEEBUBNFFU(.1SFRVJSFNFOUTBOE
2"POBEBJMZCBTJT TFSJPVTKVEHNFOUFSSPSTJO expected standards. JVT
EBUBSFDPSEJOH EBUBWFSJGJDBUJPO EBUBDPSSFDUJPO
BOEBTTPDJBUFEBDUJWJUJFTNBZPDDVS5IFTFFSSPST ARTICLE ACRONYM LISTING
BSFWFSZDSJUJDBMUPWBMJEBUJPOEPDVNFOUBUJPO CAPA Corrective Action and Preventive Action
#FDBVTFWBMJEBUJPOEPDVNFOUTBSFWFSZMJLFMZUP GMP Good Manufacturing Practice
CFSFWJFXFECZSFHVMBUPSZBVEJUPST TVCTUBOEBSE QA Quality Assurance
EPDVNFOUBUJPONBZIBWFTFSJPVTSBNJGJDBUJPOT VAC Validation Approval Committee
UPUIFPSHBOJ[BUJPOBOENBZHSFBUMZJOGMVFODFUIF
PVUDPNFPGBOBVEJU

David LeBlond
Using Probability Distributions

to Make Decisions
David LeBlond
8FMDPNF+75SFBEFSTUPi4UBUJTUJDBM7JFXQPJOUw KEY POINTS

5IJTDPMVNOQSFTFOUTTUBUJTUJDBMQSJODJQMFT 5IJTJOTUBMMNFOUPGi4UBUJTUJDBM7JFXQPJOUw
VTFGVMUPQSBDUJUJPOFSTJODPNQMJBODFBOE QSFTFOUTFJHIUFYBNQMFTDFOBSJPTUIBU
WBMJEBUJPO8FIPQFJUXJMMCFDPNFBVTFGVM EFNPOTUSBUFIPXBQSPCBCJMJUZBQQSPBDIDBO
resource. Barriers to the understanding and IFMQXJUISFTPVSDFQMBOOJOHBOEEFDJTJPO
VTFPGTUBUJTUJDBMQSJODJQMFTJODMVEFDPNQMFY NBLJOH"MMFYBNQMFTVUJMJ[FWBSJPVTEJTDSFUFPS
NBUIFNBUJDT BCTUSBDUJEFBT SFMBUJPOTIJQUPSFBM DPOUJOVPVTQSPCBCJMJUZEJTUSJCVUJPODBMDVMBUJPOT
BQQMJDBUJPOT SFRVJSFEBTTVNQUJPOT VOGBNJMJBS BWBJMBCMFJO.JDSPTPGU .4
&YDFM5IJTQSPHSBN
UFSNJOPMPHZ BOE(SFFLTZNCPMT5IJTDPMVNO JTSFBEJMZBWBJMBCMFBOEQSPWJEFTBOFBTZNFUIPE
QSFTFOUTTUBUJTUJDBMQSJODJQMFTXJUIUIFTFCBSSJFST GPSDBMDVMBUJPOT"MJTUPGSFTQFDUJWFEJTDSFUF
JONJOE*UXJMMCFPVSDIBMMFOHFUPFYQMBJO EJTUSJCVUJPOTBOEDPOUJOVPVTEJTUSJCVUJPOT
TUBUJTUJDBMDPODFQUTJOBNFBOJOHGVMXBZTPUIBU including their corresponding applications is
PVSSFBEFSTVOEFSTUBOEUIFTFJNQPSUBOUUPQJDT QSPWJEFEJO5BCMFT*BOE** SFTQFDUJWFMZ4FMFDUJPO
BOEBSFBCMFUPBQQMZTUBUJTUJDBMDPODFQUTUPXPSL PGUIFBQQSPQSJBUFEJTUSJCVUJPONPEFMGPS
situations. DBMDVMBUJPOTJTCBTFEPOUIFNPEFMTBQQMJDBUJPO
3FBEFSDPNNFOUT RVFTUJPOT BOETVHHFTUJPOT iTUPSZw5IFTFUPPMTFOBCMFRVBOUJGJBCMF
BSFOFFEFEUPIFMQVTGVMGJMMPVSPCKFDUJWFUP QSFEJDUJPOTGPSDPNNPOQIBSNBDFVUJDBM
NBLFUIJTDPMVNOVTFGVM4VHHFTUJPOTGPSGVUVSF QSPCMFNT
EJTDVTTJPOUPQJDTPSRVFTUJPOTUPCFBEESFTTFE
BSFSFRVFTUFE3FBEFSTBSFJOWJUFEUPQBSUJDJQBUF INTRODUCTION
BOEDPOUSJCVUFNBOVTDSJQUTGPSUIJTDPMVNO 5IFGJSTUJOTUBMMNFOUPGi4UBUJTUJDBM7JFXQPJOUw
$BTFTUVEJFTTIBSJOHSFHVMBUPSZTUSBUFHJFTBSF XBTQVCMJTIFEJOUIF4QSJOHJTTVFPGUIF
NPTUXFMDPNF8FOFFEZPVSIFMQUPNBLF Journal of GXP Compliance
BOEQSFTFOUFECBTJD
i4UBUJTUJDBM7JFXQPJOUwBVTFGVMSFTPVSDF QSPCBCJMJUZEJTUSJCVUJPODPODFQUT%BUBTIPVMEOPU
1MFBTFFNBJMZPVSDPNNFOUT TVHHFTUJPOT PS CFDPOTJEFSFEEFGJOJUJWF CVUUIPVHIUPGBTSBOEPN
NBOVTDSJQUTGPSQVCMJDBUJPOUPUIFNBOBHJOH WBSJBCMFTCFDBVTFPOSFQFBUFENFBTVSFNFOU UIFZ
FEJUPSBUTIBJHOFZ!BEWBOTUBSDPN FYIJCJUEJGGFSFOUWBMVFT*NBHJOFUIBUTBNQMFEBUB
BSFESBXOGSPNBIZQPUIFUJDBMQPQVMBUJPOPGBMM
[
For more Author ABOUT THE AUTHOR
information, David LeBlond, Ph.D., has 29 years experience in the pharmaceutical and medical diagnostics
go to ivthome.com/ fields. He is currently a principal research statistician supporting analytical and pharmaceutical
bios development at Abbott. David can be reached at David.LeBlond@abbott.com.

David LeBlond
Table I: Some common discrete distributions.

Distribution Excel function Distribution Story
Binomial M or P =BINOMDIST(X,,,*) A jar contains a very large (i.e., infinite) number of tablets. A known
proportion, , of the tablets are defective. We randomly select
tablets from the jar and measure X, the number in this sample that
are defective.
Bernoulli M or P =BINOMDIST(X,1,,*) Same as binomial with the sample size, = 1.
Hypergeometric M=HYPGEOMDIST(X,,K,N) K of the N tablets in a jar are known to be defective. We randomly
select tablets and measure X, the number in this sample that are
defective.
Negative binomial M=NEGBINOMIST(X,,) A lot manufacturing process has a known probability of success,
on each run. We run the process until we make good lots and
measure X, the number of failed lots produced.
Poisson M or P = POISSON(X, , *) A batch of well blended, fine, white powder is contaminated with
tiny black specks. The true mean number of black specks per unit
volume is . We randomly sample a unit volume and measure X,
the number of black specks in the sample.
*FALSE produces the probability, M, of observing exactly X (probability mass function). TRUE produces the probability, P, of observing X or less (cu-
mulative probability mass function).
QPTTJCMFWBMVFT6OEFSMZJOHQIZTJDBMQBSBNFUFST PROBING A DIFFERENCE OF

UIBUEFTDSJCFUIFEBUBHFOFSBUJPOQSPDFTTBSFLOPXO OPINIONS ABOUT PRODUCT
POMZXJUIVODFSUBJOUZBOENBZBMTPCFDPOTJEFSFE STABILITY RISK USING NORMAL PRIOR
SBOEPNWBSJBCMFT"MMSFQFBUEBUBBOEVOEFSMZJOH DISTRIBUTIONS
QBSBNFUFSTDBOCFUIPVHIUPGBTTBNQMFTGSPN %JDLBOE+BOFBSFUIFTFOJPSDIFNJTUTPOB
QSPCBCJMJUZEJTUSJCVUJPOT*ODIPPTJOHBQSPCBCJMJUZ QIBSNBDFVUJDBMEFWFMPQNFOUUFBNUIBUJTEFTJHOJOH
EJTUSJCVUJPO XFNVTUVOEFSTUBOEXIFUIFSUIFEBUB BOFXESVHQSPEVDU#PUIIBWFFYUFOTJWFFYQFSJFODF
BSFJOIFSFOUMZEJTDSFUF JF UBLFPOPOMZTQFDJGJD XJUIUIFDIFNJTUSZPGQSPEVDUTTJNJMBSUPUIFOFX
WBMVFT
PSDPOUJOVPVT JF UBLFPOJOGJOJUFMZ POFUIFZBSFEFTJHOJOH1SPHSFTTIBEHPOFXFMM
NBOZQPTTJCMFWBMVFTXJUIJOTPNFEFGJOFESBOHF
POUIFEFTJHOVOUJMUIFUFBNDBNFUPUIFJTTVFPG
%JTDSFUFEBUBTIPVMECFNPEFMFEVTJOHEJTDSFUF TUBCJMJUZ5IFQSPEVDUDBOMPTFOPNPSFUIBO
EJTUSJCVUJPOT FH CJOPNJBM
$POUJOVPVTEBUB QPUFODZVOJUTPOTUPSBHFGPSUIFEFTJSFENPOUI
TIPVMECFNPEFMFEVTJOHDPOUJOVPVTEJTUSJCVUJPOT shelf life.
FH OPSNBM
1SPCBCJMJUZEJTUSJCVUJPOTBSFBXBZ #BTFEPOIJTQSJPSLOPXMFEHFBOEFYQFSJFODF
UPVOEFSTUBOEVODFSUBJOUZBOEWBSJBUJPOBOEUIVT %JDLCFMJFWFTUIBUUIFOFXEFTJHOXJMMTVQQPSU
NBLFQSFEJDUJPOT NBOBHFSJTL BOENBLFPQUJNBM BNPOUITIFMGMJGF+BOFTQSJPSLOPXMFEHF
decisions. BOEFYQFSJFODFTVHHFTUUPIFS IPXFWFS UIBUUIF
#FGPSFUIFBEWFOUPGUIFNPEFSOQFSTPOBM OFXQSPEVDUOFFETUPIBWFBTIPSUFSTIFMGMJGF
DPNQVUFS VTFPGQSPCBCJMJUZEJTUSJCVUJPOTSFRVJSFE 3FDFOUMZ QSPHSFTTIBTTUBMMFEPWFSUIFTUBCJMJUZ
UIFVTFPGDPNQMFYUBCMFTUIBUPGUFOXFSFOPUSFBEJMZ JTTVFXJUI+BOFBOE%JDLMPDLFEJOVOQSPEVDUJWF
BWBJMBCMF.BOZDPNQMJDBUFEQSPCBCJMJUZDIBMMFOHFT EJTDVTTJPOTPWFSUIFJSEJGGFSJOHTVCKFDUJWF
TUJMMSFRVJSFUIFVTFPGTQFDJBMJ[FETUBUJTUJDBM opinions.
QBDLBHFT CVUTJNQMFQSPCBCJMJUZQSPCMFNTDBO #FGPSFQSPDFFEJOHXJUIBMFOHUIZTUBCJMJUZ
PGUFOCFTPMWFEXJUI.4&YDFM5IFPCKFDUJWFIFSF TUVEZUPSFTPMWFUIFJNQBTTF UIFQSPKFDUEJSFDUPS
JTOPUUPTVHHFTUUIBU&YDFMJTUIFUPPMPGDIPJDFGPS XBOUTUPQSPCFEFFQFSBOEMFWFSBHFUIFQSJPS
QSPCBCJMJUZBOETUBUJTUJDBMXPSL CVUUPNBLFEFDJTJPO LOPXMFEHFBOEFYQFSJFODFPGIJTUXPTFOJPSUFBN
NBLFSTBXBSFPGUIFGBDUUIBUQSPCBCJMJUZEJTUSJCVUJPO NFNCFST)FGFFMTBCFUUFSTUVEZQMBOXJMMSFTVMU
GVODUJPOTBSFMJUFSBMMZBUUIFJSGJOHFSUJQTBOEDBOCF JGUIFUFBNUBLFTBNPSFRVBOUJUBUJWFQPJOUPG
FNQMPZFEUPCFUUFSVOEFSTUBOEBOENBOBHFSJTL WJFX)FNFFUTXJUI%JDLBOE+BOFBOESFNJOET
5IFGPMMPXJOHFJHIUFYBNQMFTJUVBUJPOT UIFNPGUIFGPMMPXJOHBEWJDFGSPN-PSE,FMWJO
JMMVTUSBUFUIFVUJMJUZPGBQSPCBCJMJTUJDBQQSPBDI

1MFBTFSFNFNCFSUIBUTPNFQSPCMFNTSFRVJSF ... when you can measure what you are speaking
professional help. It is always wise to consult a about and express it in numbers, you know
USBJOFETUBUJTUJDJBOSFHBSEJOHBOZDPNQMFYSJTL something about it, but when you cannot ... express
BTTFTTNFOUQSPCMFN PSXIFOUIFWBMJEJUZPGDFSUBJO it in numbers, your knowledge is of a meager and
BTTVNQUJPOTJTVODMFBS unsatisfactory kind ...

David LeBlond
Table II: Some common continuous distributions.

Distribution Excel function Distribution Story
Normal or D or P =NORMDIST(X,,,*) In the manufacture of a drug lot, an amount of API exactly needed
Gaussian to achieve the desired potency, , is added. However a large (i.e.,
X = NORMINV(P,,) infinite) number of subsequent processing steps (e.g., weighing and
= + *NORMINV(P,,1) adding various excipients) each produce small, random increases
or decreases in final lot potency. The net effect produces a stan-
dard deviation of about the mean of . Make a lot and measure
the observed potency X.
Lognormal P=LOGNORMDIST(X,,) The measured response, X, of an analytical test is the multiplicative
product of the effects of many small, independent factors (volumes,
X=LOGINV(P,,) path-length, etc). The natural log of X follows a normal distribution
with mean and standard deviation .
Exponential D or P=EXPONDIST(X,,*) An electrical supply system experiences occasional random black-
outs. The long term mean blackout frequency is known to be con-
X = -LN(1-P)/ stant, . Measure X, the time between blackouts.
Gamma D or P =GAMMADIST(X,,1/,*) Same as the Exponential above, except measure X, the time re-
quired for 1 blackouts to occur.
X=GAMMAINV(P,,1/)
Weibull D or P =WEIBULL(X,,1/,*) Same as the Exponential above except the long term mean black-
out frequency is not constant but increases or decreases over time
X = =(-LN(1-P)/)^(1/) according to the parameter >0. Measure X, the time between
blackouts.
Beta P = BETADIST(X,,+1,0,1) A very large lot of tablets is known to contain defects. Tablets are
sampled randomly from the lot until defective tablets have been
X = BETAINV(P,,+1,0,1) found. Measure X, the fraction of the lot tested.
Rectangular or P=X Same as the Beta, with = = 1.
Uniform D=1
Chi-squared P = 1-CHIDIST(X,1) A sample of size is taken from a normal random variable with
known standard deviation = . The sample variance (V=square of
D = GAMMADIST(X,(1)/2,2,TRUE) the standard deviation) is calculated. Measure X=(1)V/ 2.
X = CHIINV(1P,1)
Students-t if X<0 A sample of size is taken from a normal population with
P = TDIST(ABS(X),1,1) mean = 0. The sample average, X , and standard deviation, s, are
calculated. X
if X0 Measure X = s/ .
P = 1-TDIST(X,1,1)
if 0.5 P < 1
X = TINV(2*(1-P),1)
if 0 < P < 0.5

X = -TINV(2*P, 1)
F P = 1-FDIST(X,1, 1) A sample of size is taken from a normal population and the vari-
ance (V1) is calculated. A second sample of size is also taken
X = FINV(1 - P,1,1) from the same population and the variance (V2) is calculated. Mea-
sure X = V1/V2 .
*FALSE produces the probability of density, D, of observing exactly X. TRUE produces the probability, P, of observing x or less (cumulative probabil-
ity density function).
#PUIDIFNJTUTBHSFFUIBUUIFBDUJWF UIFSFGPSF TVHHFTUTUIBUUIFZFYQSFTTUIFJS

DPNQPOFOUJTTFOTJUJWFUPPYJEBUJPOPWFSUJNF PQJOJPOTBCPVUUIFJOTUBCJMJUZJOUIFGPSNPGB
BOEBTPMVUJPOJTUPJODMVEFBOBOUJPYJEBOU DPOUJOVPVTOPSNBMQSPCBCJMJUZEJTUSJCVUJPO TFF
JOUIFGPSNVMBUJPOIPXFWFS UIJTNFBTVSF UIFOPSNBMEJTUSJCVUJPOTUPS ZJO5BCMF**
)F
JTOPUDPNQMFUFMZFGGFDUJWFBOEUIFEFHSFFPG BTLTFBDIDIFNJTUUPQSPWJEFUIFJSCFTUFTUJNBUF
QSPUFDUJPOTFFNTUPWBS ZGSPNMPUUPMPU#PUI PG
5IFFYQFDUFENFBOQPUFODZMPTTBGUFS
DIFNJTUTBHSFFUIBUUIFWBSJBCJMJUZJTEVFUP NPOUITTUPSBHF BOE
5IFFYQFDUFEUI
NBOZTNBMMBEEJUJWFFGGFDUT5IFEJSFDUPS QFSDFOUJMFPGUIFNFBOQPUFODZMPTT

David LeBlond
"GUFSDBSFGVMUIPVHIU CPUIDIFNJTUTBHSFFUIBU
Figure 1: Estimates of the mean and standard deviations
POBWFSBHF QSPEVDUMPUTTIPVMEMPTFBCPVUVOJUT
using MS Excel function NORMDIST.
PGUIFJSQPUFODZBGUFSNPOUITPGTUPSBHF#VU
UIFZEJTBHSFFPOUIFUIQFSDFOUJMF+BOFGFFMT 0.8 Dicks prior
UIBUPGMPUTXPVMEMPTFVOJUTPGQPUFODZPS Janes prior
Probability density
MFTT XIJMF%JDLGFFMTUIBUPGMPUTXPVMEMPTF
0.6
OPNPSFUIBOVOJUTPGQPUFODZ5IFEJSFDUPS
UBLFTVOJUTBTUIFOPSNBMNFBO BOEVTFTUIF
following relationship 0.4
/03.*/7 1 ) = + /03.*/7 1
0.2
UPFTUJNBUFUIFOPSNBMTUBOEBSEEFWJBUJPOTJO
0
&YDFMBTGPMMPXT
0 2 4 6 8 10
'PS+BOF = (9
/03.*/7
=
True potency loss on 24 month storage
'PS%JDL =
/03.*/7

"SNFEXJUIUIFTFQSJPSFTUJNBUFTPGUIFNFBO PREDICTING CONTROL CHART

BOETUBOEBSEEFWJBUJPOT UIFNBOBHFSHFOFSBUFT TREND TEST DECISION ERROR RATES
'JHVSFVTJOHUIF&YDFMGVODUJPO/03.%*45 -PTT USING THE NORMAL AND BINOMIAL
'"-4&
)FFYQMBJOTUIBUUIFTFEJTUSJCVUJPOT DISTRIBUTIONS
DBQUVSFUIFQSJPSFYQFDUBUJPOTPGFBDIDIFNJTU 5IF8FTUFSO&MFDUSJD)BOECPPLTVHHFTUTBTFUPG
RVBOUJUBUJWFMZ*UCFDPNFTDMFBSUIBUUIFEJGGFSFODF EFDJTJPOSVMFTGPSEFUFDUJOHOPOSBOEPNQBUUFSOT
PGPQJOJPOJTEVFUPPQJOJPOTBCPVUUIFNBHOJUVEF JODPOUSPMDIBSUT
4VDIDPOUSPMDIBSUSVMFT
PGMPUUPMPUWBSJBUJPO XJUI+BOFTQSJPSEJTUSJCVUJPO BDUMJLFTNPLFBMBSNT5IFZBMFSUVTUPQPTTJCMF
TIPXJOHHSFBUFSTQSFBEUIBO%JDLT DIBOHFTPSESJGUTJOPVSNFUIPETBOEQSPDFTTFTUIF
TPDBMMFEiTQFDJBMDBVTFTwUIBUDBOCFJOWFTUJHBUFE
5PFYQSFTTUIFFYQFDUFEGBJMVSFSBUFRVBOUJUBUJWFMZ 5IFZQSFWFOUVTGSPNSFTQPOEJOHUPUIFiXIJUF
UIFEJSFDUPSVTFTUIFFRVBUJPO OPJTF wBMTPLOPXOBTiDPNNPODBVTFwWBSJBUJPO
UIBUJTBMXBZTQSFTFOUCVUGPSNJEBCMFUPJOWFTUJHBUF
SPCBCJMJUZUIBUPCTFSWFEMPTTJTHSFBUFSUIBO
1 )PXFWFS FWFODPOUSPMDIBSUSVMFTBSFJNQFSGFDU
units = 1 /03.%*45 536&
8IBUBSFUIFJSGBMTFQPTJUJWFBMBSNSBUFT *GXFEFDJEF
UPVTFUIFTFSVMFT IPXPGUFODBOXFFYQFDUUPTUBSU
JO&YDFMUPFTUJNBUFUIFQSJPSQSPCBCJMJUZPG BOJOWFTUJHBUJPOXJUIMJUUMFIPQFPGVODPWFSJOHB
NBLJOHBMPUUIBUDPVMEOPUNFFUUIFTUBCJMJUZ TQFDJBMDBVTF 8FDBOBEESFTTUIFTFRVFTUJPOTCZ
SFRVJSFNFOUPGMFTTUIBOVOJUTMPTTPOTUPSBHF VTJOHTPNFTJNQMFQSPCBCJMJUZEJTUSJCVUJPOUPPMT
5IFSFTVMUJOHQSJPSGBJMVSFSBUFTGPS%JDLBOE+BOFBSF 'PVSDPNNPOSVMFTBSFJMMVTUSBUFEJO'JHVSF5P
BOE SFTQFDUJWFMZ$MFBSMZ CBTFE BQQMZUIFTFSVMFT EJWJEFUIFDIBSUJOUPUISFF[POFT
POIFSQSJPSCFMJFGT +BOFTDPODFSOJTXBSSBOUFE4IF " # BOE$UIBUEFOPUFUIFEJTUBODFJOTUBOEBSE
QSFEJDUFEUIBUPGGVUVSFMPUTDPVMEOPUNFFU EFWJBUJPO
VOJUTGSPNUIFNFBO
*OQSBDUJDF
UIFTUBCJMJUZTQFDJGJDBUJPO5IFDIFNJTUTBHSFFUIBU and XPVMECFFTUJNBUFEGSPNUIFDPOUSPMDIBSU
UJHIUFSDPOUSPMPWFSUIFMFWFMPGBOUJPYJEBOUXPVME EBUB*OFTUJNBUJOHUIFGBMTFBMBSNSBUFGPSFBDISVMF
CFOFFEFE XFXJMMBTTVNFUIBU and are known precisely.
8IJMFTUBUJTUJDTBMPOFDBOOPUSFTPMWFB 8FXJMMBMTPBTTVNFUIBUUIFSFBSFOPTQFDJBMDBVTFT
DIFNJDBMEJMFNNB UIJTFYBNQMFEFNPOTUSBUFT QSFTFOUBOEUIBUUIFDPNNPODBVTFWBSJBUJPOJT
IPXUIFVTFPGQSPCBCJMJUZUPPMTDBOGPDVT EJTUSJCVUFESBOEPNMZBOEGPMMPXTBDPOUJOVPVTOPSNBM
EJTDVTTJPOCZGPSDJOHRVBOUJUBUJWFUIJOLJOH EJTUSJCVUJPO TFF5BCMF**
5IFSVMFTBQQMZUPPOMZPOF
#ZSFFYQSFTTJOHBWFSCBMEJTBHSFFNFOUBTB TJEFPGUIFDFOUFSMJOFBUBUJNF8FTUBUFUIFSVMFTBT
EJGGFSFODFJOQSPCBCJMJUZEJTUSJCVUJPOT MPUUPMPU follows:
WBSJBUJPOXBTJEFOUJGJFEBTUIFSFBMJTTVF5IJT 0OFQPJOUCFZPOE[POF"
SFDPHOJUJPONJHIUMFBEUPGSVJUGVMEJTDVTTJPOT 5XPPVUPGUISFFDPOTFDVUJWFQPJOUTCFZPOE
BCPVUQSPDFTTDBQBCJMJUZJNQSPWFNFOUUIBUDPVME [POF#
ZJFMEBWJBCMFQSPEVDU 'PVSPVUPGGJWFDPOTFDVUJWFQPJOUTCFZPOE
[POF$

David LeBlond
Figure 2: Four common control chart rules. #*/0.%*45 '"-4&

=
*UDBOCFTFFOUIBUUIFSVMFTIBWFCFFOEFTJHOFE
1 TPUIBUUIFPOFTJEFEGBMTFBMBSNSBUFGPSFBDI
3 JTCFMPX UJNFTJOBUIPVTBOETFQBSBUF
A
2 4 TFUTPGQPJOUTFYBNJOFE
"TTUBUFEXJUISVMF
B 2 JUNVTUCFLFQUJONJOEUIBUGBJMVSFTDBOPDDVS
1
C POFJUIFSTJEFPGUIFDFOUFSMJOF TPUIFUXPTJEFE
GBMTFBMBSNSBUFTBSFEPVCMFUIFWBMVFTDBMDVMBUFE
C BCPWF"OPUIFSJNQPSUBOUDPOTJEFSBUJPOJTUIF
1 3
B QSPCMFNPGNVMUJQMJDJUZ8IFONVMUJQMFTUBUJTUJDBM
2 UFTUTBSFBDUJWFMZFNQMPZFE UIFGBMTFBMBSNSBUF
A UIBUPOFPSNPSFXJMMBDDJEFOUBMMZiTPVOEwNBZCF
3
correspondingly higher. If the tests are independent
XJUIWFSZMPXGBMTFBMBSNSBUFT UIFVTFPGUFTUT
NBZJODSFBTFUIFPWFSBMMGBMTFBMBSNSBUFOFBSMZ
GPME5IVTDBVUJPOJTBEWJTFEXIFOFNQMPZJOH
&JHIUDPOTFDVUJWFQPJOUTPOUIFTBNFTJEFPG
NVMUJQMFDPOUSPMDIBSUSVMFTUPHFUIFS
the center line.
5IFGBMTFBMBSNSBUFGPSSVMFJTTJNQMZUIF PREDICTING IN-PROCESS CONTENT
QSPCBCJMJUZPGPCTFSWJOHBTJOHMFQPJOUBUMFBTU UNIFORMITY FAILURE RISK USING THE
GSPNUIFNFBO4JODFUIFDPOUJOVPVTOPSNBM NORMAL AND BINOMIAL DISTRIBUTIONS
EJTUSJCVUJPOJTTZNNFUSJDBM XFDBOGPDVTPOUIF "QBSUJDVMBSOFXQIBSNBDFVUJDBMUBCMFUQSPEVDUDBO
TJEFCFMPXUIFNFBO5IJTQSPCBCJMJUZDBOCF CFNBOVGBDUVSFETPUIBU POBWFSBHF FBDICBUDI
PCUBJOFEVTJOHUIFGPMMPXJOH&YDFMGPSNVMB DPOUBJOTNHPGBDUJWFESVH)PXFWFS EVFUPUIF
TNBMMBEEJUJWFFGGFDUTPGNBOZQSPDFTTWBSJBCMFT
VMFGBMTFBMBSNSBUF=/03.%*45
3 UIFSFJTXJUIJOCBUDI UBCMFUUPUBCMFUWBSJBUJPO
536&
= 5IFXJUIJOCBUDITUBOEBSEEFWJBUJPOJOUBCMFUESVH
5IJTNFBOTUIBUSVMFDBOCFFYQFDUFEUPGBJMPO DPOUFOUJTSFQSPEVDJCMZNH
PVUPGDPOUSPMDIBSUQPJOUTPSPGUIF &BDICBUDIPGUBCMFUTNVTUQBTTBOJOQSPDFTT
UJNF8FIBWFUPLFFQJONJOEUIBUUIFSVMFNJHIU VOJUEPTFDPOUFOUVOJGPSNJUZUFTU5IFUFTUSFRVJSFT
also fail on the upper side of the chart and so the BQPUFODZBTTBZPOFBDIPGSBOEPNMZTFMFDUFE
DPNCJOFEGBMTFBMBSNSBUFJT UBCMFUTGSPNUIFCBUDI5IFQPUFODZBTTBZJTWFSZ
3VMFJTBMJUUMFNPSFDPNQMJDBUFE'JSTUOPUFUIBU BDDVSBUFBOEQSFDJTF5IFUFTUGBJMTBOEUIFCBUDI
UIFQSPCBCJMJUZPGFYDFFEJOH on the lower side NVTUCFEJTDBSEFEJGUIFNFBTVSFEQPUFODZPGBOZ
PGUIFDIBSUDBOCFPCUBJOFEVTJOH/03.%*45 PGUIFUBCMFUTFYDFFETUIFSBOHFoNH
536&
/FYUOPUFUIBUUIFGBMTFBMBSN 'BJMVSFNFBOTUIBUWBMVBCMFESVHQSPEVDUJTMPTUBOE
SBUFXJMMCFFRVBMUPUIFCJOPNJBMQSPCBCJMJUZPG NBOVGBDUVSJOHFGGJDJFODZJTSFEVDFE'PSQMBOOJOH
SBOEPNMZTFMFDUJOHPVUPGEFGFDUJWFQPJOUTGSPN BOEGJOBODJOHQVSQPTFT PQFSBUJPOTOFFETUPLOPX
BMBSHFQPQVMBUJPOPGQPJOUTJOXIJDIPGUIF IPXPGUFOGVUVSFCBUDIFTXJMMGBJMUIFJOQSPDFTT
QPJOUTBSFiEFGFDUJWFw4FFUIFiTUPSZwBTTPDJBUFEXJUI DPOUFOUVOJGPSNJUZ
UIFDBUFHPSJDBMCJOPNJBMEJTUSJCVUJPOJO5BCMF*5IVT 5IFDIJFGFOHJOFFSLOPXTUIBU BTTVNJOHUIF
UIFSFRVJSFEQSPCBCJMJUZDBOCFPCUBJOFEBTGPMMPXT QSPDFTTJTCFIBWJOHBTFYQFDUFE UBCMFUQPUFODJFT
TIPVMECFEJTUSJCVUFEBDDPSEJOHUPBDPOUJOVPVT
VMFGBMTFBMBSNSBUF=
3 OPSNBMEJTUSJCVUJPO TFF5BCMF**
XJUIBNFBO
#*/0.%*45 '"-4&
= PGNHBOETUBOEBSEEFWJBUJPO PGNH5IVT
UIFQSPCBCJMJUZUIBUBSBOEPNUBCMFUXJMMFYDFFEUIF
5IFFSSPSQSPCBCJMJUJFTGPSSVMFTBOENBZCF oNHSBOHFDBOCFDBMDVMBUFEBTGPMMPXT
PCUBJOFETJNJMBSMZBTGPMMPXT
3VMFGBMTFBMBSNSBUF= /03.%*45 536&

+1
#*/0.%*45 /03.%*45 536&
=
/03.%*45 536&
'"-4&
=
4JNJMBSUPUIFDPOUSPMDIBSUFYBNQMFHJWFOBCPWF
3VMFGBMTFBMBSNSBUF= UIFQSPCBCJMJUZPGPCUBJOJOHPSNPSFPVUPGSBOHF

David LeBlond
UBCMFUTJOBSBOEPNTBNQMFPGXJMMGPMMPXB
Figure 3: Scaled probability density plots.
DBUFHPSJDBMCJOPNJBMEJTUSJCVUJPO TFF5BCMF*
BOE
XFDBOPCUBJOUIFCBUDIGBJMVSFSBUFBT
0.18
1 #*/0.%*45 536&

= 0.16
True SD=0.5
0.14
Probability density
True SD=0.7
5IVTGVUVSFCBUDIFTDBOCFFYQFDUFEUPGBJMUIF 0.12
JOQSPDFTTDPOUFOUVOJGPSNJUZUFTUPGUIF
0.1
UJNF
0.08
ASSESSING THE SUITABILITY OF 0.06
ANALYTICAL METHODS USING THE CHI- 0.04
SQUARED DISTRIBUTION
0.02
*OUIJTTDFOBSJP UXPIJHIQSFTTVSFMJRVJE
DISPNBUPHSBQIZ )1-$
BOBMZUJDBMNFUIPETBSF 0
CFJOHEFWFMPQFE*UJTEFDJEFEUIBUTJYDPOUSPM 0 0.5 1.0 1.5 2.0
TBNQMFTXJMMCFSVOXJUIFBDIBOBMZUJDBMSVO5P
Observed variance
NFFUUIFTZTUFNTVJUBCJMJUZSFRVJSFNFOUT JF GPS
UIFSFTVMUTPGVOLOPXOTBNQMFTJOUIBUSVOUPCF
SFQPSUBCMF
UIFPCTFSWFETUBOEBSEEFWJBUJPOPG
UIFTFOSFQMJDBUFUFTUTNVTUCFOPNPSFUIBO EPFTOPUTVQQMZB$IJTRVBSFEEFOTJUZGVODUJPO
PGUIFOPNJOBMDPOUSPMDPODFOUSBUJPO"TZTUFN 4VDIGVODUJPOTBSFBWBJMBCMFJOTPGUXBSFTVDI
TVJUBCJMJUZGBJMVSFJTDPTUMZJOUFSNTPGBOBMZTU BT4"4 +.1 PS3%JSFDUDBMDVMBUJPOPGUIF$IJ
UJNFBOESFBHFOUT'VSUIFS BOBMZUJDBMFRVJQNFOU TRVBSFEEFOTJUZJTQPTTJCMFCVURVJUFDPNQMJDBUFE
JTMJNJUFE$POTFRVFOUJBMMZ QSPQFSSFTPVSDF 'PSUVOBUFMZ UIF$IJTRVBSFEEJTUSJCVUJPOJTB
NBOBHFNFOUSFRVJSFTUIFBCJMJUZUPBOUJDJQBUFUIF TQFDJBMDBTFPGUIFDPOUJOVPVT(BNNBEJTUSJCVUJPO
SBUFPGTZTUFNTVJUBCJMJUZGBJMVSFT TFF5BCMF**
TVDIUIBU
5IFUXPBOBMZUJDBMNFUIPETBSFWFSZBDDVSBUF
CVUIBWFXJUIJOSVOTUBOEBSEEFWJBUJPOT ) of IJTRVBSFEQSPCBCJMJUZEFOTJUZ
$
BOE SFTQFDUJWFMZ*OCPUIDBTFT UIF =(".."%*45
7
536&
XJUIJOSVOEFWJBUJPOTGSPNUIFFYQFDUFEDPOUSPM
WBMVFBSFEJTUSJCVUFEOPSNBMMZ5IJTTVHHFTUTUIBU 5IJTFRVBUJPOXBTVTFEUPHFOFSBUFUIFTDBMFE
BDPOUJOVPVT$IJTRVBSFEEJTUSJCVUJPONJHIUCF QSPCBCJMJUZEFOTJUZQMPUTTIPXOJO'JHVSF XJUI7
VTFEUPEFTDSJCFUIFEJTUSJCVUJPOPGUIFXJUIJOSVO QMPUUFEPOUIFIPSJ[POUBMBYJT5IFFYDFTTJWFGBJMVSF
WBSJBODF JF 7 UIFTRVBSFPGUIFPCTFSWFETUBOEBSE rate with the second assay is clearly due to the fact
EFWJBUJPO
UIBUUIF$IJTRVBSFEEFOTJUZUFOETUPIBWFBMPOH
5IF$IJTRVBSFEiTUPSZwEFTDSJCFTTBNQMJOHGSPN SJHIUUBJM JF JTSJHIUTLFXFE
XIFOJTTNBMM
BOPSNBMQPQVMBUJPOXIPTFTUBOEBSEEFWJBUJPOJT
TFF5BCMF**
5IJTQBSBNFUFSDPSSFTQPOETUPUIF SETTING A PRECISION SPECIFICATION
XJUIJOTUBOEBSEEFWJBUJPOTPGPS5IVT LIMIT FOR AN ANALYTICAL METHOD
UIFQSPCBCJMJUZPGFYDFFEJOHUIFMJNJUDBOCF TRANSFER USING THE F DISTRIBUTION
DBMDVMBUFEGPSUIFUXPBOBMZUJDBMNFUIPETBTGPMMPXT 8IFOBOBOBMZUJDBMNFUIPEJTUSBOTGFSSFEGSPN
'BJMVSFSBUFGPSGJSTUNFUIPE POFMBCPSBUPSZUPBOPUIFS JUTVTFJOUIFTFDPOE
=$)*%*45
?? 1) = MBCPSBUPSZNVTUGJSTUCFWBMJEBUFE5IJTJTUPBTTVSF
UIBUEJGGFSFOUBOBMZTUT VTJOHEJGGFSFOUFRVJQNFOU
' BJMVSFSBUFGPSTFDPOENFUIPE BOESFBHFOUT BOEQPTTJCMZTMJHIUMZEJGGFSFOU
=$)*%*45
??
QSPDFEVSFT
DBOBDIJFWFUIFTBNFBOBMZUJDBM
QFSGPSNBODF"LFZFMFNFOUPGQFSGPSNBODF
4VSQSJTJOHMZ UIFTNBMMJODSFBTFJOUSVFTUBOEBSE GPSRVBOUJUBUJWFBTTBZTJTQSFDJTJPOBTNFBTVSFE
EFWJBUJPOGSPNUPSFTVMUTJOOFBSMZB CZNFUIPETUBOEBSEEFWJBUJPO0OFHPBMPGB
GPMEJODSFBTFJOTZTUFNTVJUBCJMJUZGBJMVSFSBUF NFUIPEUSBOTGFSQSPUPDPMJT UIFSFGPSF UPNBLFB
5IFTFDPOEBTTBZXPVMEGBJMOFBSMZPVUPGFWFSZ TUBUJTUJDBMDPNQBSJTPOPGTUBOEBSEEFWJBUJPOT5IF
BOBMZUJDBMSVOT QFSGPSNBODFNFUSJDNJHIUCFBSBUJPPGUIFUXP
*UJTPGUFOJOTUSVDUJWFUPBDUVBMMZQMPUUIF PCTFSWFETUBOEBSEEFWJBUJPOT UFTUDPOUSPM

QSPCBCJMJUZEFOTJUZGVODUJPOUPTFFUIFTIBQFPG *GUIJTQFSGPSNBODFNFUSJDJTTRVBSFE UIF
UIFTBNQMJOHEJTUSJCVUJPO6OGPSUVOBUFMZ &YDFM SFTVMUJOHTUBUJTUJDJTBSBUJPPGUXPWBSJBODFT5IJT

David LeBlond
Figure 4: Cumulative F distribution. XBOUTUPCFBUMFBTUTVSFUIBUBGBJMVSFXPVME

OPUPDDVSCZTJNQMFSBOEPNDIBODF POFTIPVME
DPOTJEFSBDDFQUBODFMJNJUTGPSUIF'SBUJPBCPWF
1
4QFDJGJDBMMZJUDBOCFTIPXOUIBU
0.9
Cumulative probability density
0.8 1'%*45 1) =

0.7
*GPOFVTFTBOBDDFQUBODFMJNJUPG
0.6 DPSSFTQPOEJOHUPBTUBOEBSEEFWJBUJPOSBUJPPG
0.5 UIFQSPUPDPMDSJUFSJPOGPSQSFDJTJPOXJMMIBWFB
GBMTFBMBSNSBUFPGMFTTUIBOCFDBVTFPGSBOEPN
0.4
DIBODFBMPOF5IFBDDFQUBODFDSJUFSJPODBOCF
0.3 SFEVDFECZJODSFBTJOHUIFOVNCFSPGSFQMJDBUFTGPS
0.2 FBDIMBCPSBUPSZSVO5IVTUIF'EJTUSJCVUJPODBOCF
VTFEUPTFMFDUCPUIUIFSFRVJSFETBNQMFTJ[FBOEUIF
0.1
decision criterion.
0 /PUFUIBUUIF'UFTUEFTDSJCFEBCPWFJTBPOF
0 2 4 6 TJEFEUFTU8FIBWFDMFBSMZEFGJOFEPVS'SBUJP
Observed variance ratio UFTUDPOUSPM
BOEXFPOMZBSFDPODFSOFEBCPVU
(test/control) GBJMVSFTPOUIFIJHITJEF MBSHF'WBMVFT
*GXF
XFSFDPODFSOFEBCPVUUFTUJOHGPSFRVBMJUZPGUIF
WBSJBODFTXFXPVMEBMTPCFDPODFSOFEXJUIGBJMVSFT
POUIFMPXTJEF TNBMM'WBMVFT
*OUIFMBUUFSDBTF
TUBUJTUJDJTEFOPUFE'BGUFSUIF&OHMJTITDJFOUJTU4JS XFXPVMEOFFEUPDPOTUSVDUBUXPTJEFEUFTUUP
3POBME'JTIFSXIPGJSTUUBCVMBUFEJUTTBNQMJOH detect significant differences of either kind.
EJTUSJCVUJPO*GUIFWBSJBODFTQSPEVDFECZUIFUFTU
BOEDPOUSPMMBCPSBUPSJFTBSFJOGBDUUSVMZFRVBMBOE SETTING A SPECIFICATION LIMIT FOR
JGUIFWBSJBUJPOPGNFBTVSFNFOUTJTEJTUSJCVUFE A POWDER BLEND VALIDATION USING
OPSNBMMZ UIFOUIFTJUVBUJPONBUDIFTUIF THE POISSON DISTRIBUTION
'EJTUSJCVUJPOiTUPSZw TFF5BCMF**
BOEXFNBZVTF "QPXEFSCMFOEJOHQSPDFTTNVTUCFWBMJEBUFEUP
UIF'EJTUSJCVUJPOUPTFUBTUBUJTUJDBMBDDFQUBODFMJNJU BTTVSFUIBUUIFQSPDFTTEJTUSJCVUFTUIFJOHSFEJFOUT
GPSUIFQSFDJTJPODSJUFSJPOJOUIFNFUIPEUSBOTGFS FWFOMZUISPVHIPVUUIFCBUDI5IFCMFOEDPOTJTUT
protocol. of two white ingredients whose particles are of
$POTJEFSBWFSZTJNQMFQSPUPDPMGPSBTTFTTNFOU FRVBM XFMMDPOUSPMMFETJ[FBOEIBWFWFSZTJNJMBS
PGQSFDJTJPO&BDIMBCPSBUPSZUFTUTSFQMJDBUFT QIZTJDBMQSPQFSUJFT"OFOHJOFFSIBTEFDJEFEUP
PGBDFSUBJODPOUSPMBOEEFUFSNJOFTUIFWBSJBODF VTFIJTOFXGPVOEVOEFSTUBOEJOHPGQSPCBCJMJUZ
PGUIFTFSFQMJDBUFT5IFOUIF'WBSJBODFSBUJPJT EJTUSJCVUJPOTUPTJNQMJGZUIFWBMJEBUJPOUSJBMT)F
DBMDVMBUFE UFTUDPOUSPM
5IFQSFDJTJPODSJUFSJBQBTT QMBOTUPJODMVEFBUSBDFBNPVOUPGBUIJSEJOFSU
BTMPOHBTUIFSBUJPJTBUPSCFMPXTPNFMJNJU*GUIF JOHSFEJFOU BMTPPGFRVBMQBSUJDMFTJ[FBOEQIZTJDBM
MJNJUJTTFUUPPMPX UIFDSJUFSJPOXPVMEGBJMUPP QSPQFSUJFT FYDFQUUIBUUIFQBSUJDMFTBSFCMBDL5IFZ
PGUFOTJNQMZCZSBOEPNDIBODF5IFMJNJUTIPVME DBOCFFBTJMZJEFOUJGJFEBOEDPVOUFEJOBHSBNPG
CFTFUIJHIFOPVHI CVUOPUUPPIJHI
TVDIUIBU QPXEFSCZFZFVTJOHBNBHOJGZJOHHMBTT
BQSPUPDPMGBJMVSFHJWFTBCPOBGJEFJOEJDBUJPOPG 5IFFOHJOFFSSFBTPOTUIBUJGUIFCMFOEJOHJT
QSFDJTJPOJOGFSJPSJUZJOUIFUFTUMBCPSBUPSZ8IBUXF FWFO UIFEJTUSJCVUJPOPGCMBDLiTQFDLTwJOUIF
SFBMMZXBOUUPLOPXJTUIFQSPCBCJMJUZPGPCUBJOJOH GJOBMCMFOETIPVMEGPMMPXBDBUFHPSJDBM1PJTTPO
BO'SBUJPCFMPXTPNFMJNJU*OPUIFSXPSET XF EJTUSJCVUJPO TFFUIF1PJTTPOiTUPSZwJO5BCMF
XBOUUPLOPXUIFDVNVMBUJWF'EJTUSJCVUJPO5IF *
)FQMBOTUPJOUSPEVDFFOPVHIPGUIFUSBDF
DVNVMBUJWF'EJTUSJCVUJPOGPSBSBUJPPGWBSJBODFT JOHSFEJFOUTPUIBUBHSBNPGUIFGJOBMQPXEFS
FBDIPCUBJOFEGSPNSFQMJDBUFTPGJTHJWFOBT XJMMDPOUBJO POBWFSBHF CMBDLTQFDLT5IF
follows: UFTUGPSFBDICBUDIXJMMDPOTJTUPGDBSFGVMMZ
SFNPWJOHFYBDUMZHSBNPGQPXEFSGSPNB
SPCBCJMJUZUIBUPCTFSWFE'SBUJPJT9
1 SBOEPNMPDBUJPOJOUIFCMFOEFECFEVTJOHB
1'%*45 9 1) UIJFG BOEDPVOUJOHUIFOVNCFSPGCMBDLTQFDLT
PCTFSWFEJOUIFTBNQMF*GNVDIGFXFSPSNVDI
5IJTDVNVMBUJWFEJTUSJCVUJPOJTTIPXOJO'JHVSF HSFBUFSUIBOTQFDLTBSFTFFO UIFCBUDIXJMMCF
XJUIUIFSBUJP'HJWFOPOUIFIPSJ[POUBMBYJT*GPOF KVEHFEUPGBJMUIFCMFOEVOJGPSNJUZWBMJEBUJPO

David LeBlond
Figure 5: Probability mass function.
0.100
0.090
Probability mass
0.080
0.070
0.060
0.050
0.040
0.020
0.010
0.000
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Observed number of specs per sample
5IVTUIFUFTUXJMMCFUXPTJEFE5IFRVFTUJPOJT Figure 6: Cumulative distribution function of t.

XIBUMPXFSBOEVQQFSMJNJUTTIPVMECFQMBDFE
POUIFCMBDLTQFDLDPVOU5IFMJNJUTIPVMECF 1
TFUXJEFFOPVHIUPBWPJEGBJMJOHUIFDSJUFSJPO
CZSBOEPNDIBODFBMPOF ZFUOPUTPXJEFUIBU 0.9
Cumulative probability density
JNQPSUBOUMFWFMTPGTUSBUJGJDBUJPOJOUIFCMFOEBSF 0.8
NJTTFE*ODIPPTJOHUIFMJNJUT JUDBOCFIFMQGVM
UPMPPLBUUIFFYQFDUFE1PJTTPOEJTUSJCVUJPOGPS 0.7
UIJTTJUVBUJPO'SPN5BCMF* UIFQSPCBCJMJUZNBTT 0.6
GVODUJPOJTPCUBJOFEBTGPMMPXT
1 SPCBCJMJUZPGPCTFSWJOHFYBDUMZ9CMBDLTQFDLT= 0.5
10*440/ 9 '"-4&
0.4
5IFQSPCBCJMJUZNBTTGVODUJPOJTTIPXOJO'JHVSF 0.3
XJUI9QMPUUFEPOUIFIPSJ[POUBMBYJT*UXPVME 0.2
BQQFBSUIBUOPSNBMSBOEPNWBSJBUJPODPVMEFBTJMZ
QSPEVDFBSBOHFPG9WBMVFTCFUXFFOBOE 0.1
5IFFOHJOFFSGFFMTUIBUUIFUFTUTIPVMEOPUGBJMCZ 0
DIBODFNPSFUIBOPGUIFUJNF)FDBMDVMBUFT -4 -3 -2 -1 0 1 2 3 4
GBJMVSFSBUFTPOUIFMPXBOEIJHITJEFTSFTQFDUJWFMZ
as follows: Observed t statistic
SPCBCJMJUZUIBU9<=10*440/

1 FYJTUJOH DPOUSPM
CSBOE5IFUSVF QPQVMBUJPO

536&
= CBMBODFUPCBMBODFTUBOEBSEEFWJBUJPODBOCF
BTTVNFEFRVBMGPSCPUICSBOETXJUIOPSNBMFSSPST
1 SPCBCJMJUZUIBU9>=10*440/ BMUIPVHIUIFUSVFNFBOTBOETUBOEBSEEFWJBUJPO
536&
= are unknown. Ten (=/
CBMBODFTPGFBDICSBOE
XJMMCFUFTUFECZXFJHIJOHUIFTBNFTUBOEBSE
5IVTBOBDDFQUBODFDSJUFSJPOPGUPJODMVTJWF PODFPOFBDICBMBODFCZUIFTBNFPQFSBUPS5IF
XPVMETBUJTGZUIFSFRVJSFEGBMTFBMBSNSBUFPG CBMBODFTXJMMCFUFTUFEJOSBOEPNPSEFS5IF
PCTFSWFE TBNQMF
NFBO X ) and standard
SETTING A PROCESS MEAN EFWJBUJPO 4
PGFBDICSBOEXJMMCFDBMDVMBUFE
SPECIFICATION IN A COMPARABILITY *OUIFQSFWJPVTFYBNQMFT BTUBUJTUJD XIPTF
STUDY USING THE T-DISTRIBUTION TBNQMJOHEJTUSJCVUJPOXBTDPOTUSVDUFEVOEFSUIF
5IFSFMBUJWFBDDVSBDZPGBOFX UFTU
CSBOEPG BTTVNQUJPOUIBUUIFQSPDFTTFTCFJOHDPNQBSFE
MBCPSBUPSZCBMBODFJTCFJOHDPNQBSFEUPUIF XFSFUIFTBNF XBTLOPXOUIFOUIFBDDFQUBODF

David LeBlond
Figure 7: Population as the area of a unit square. 5%*45 9 +

GPS9>
5IFDVNVMBUJWFEJTUSJCVUJPOGVODUJPOPGUIJT
TUBUJTUJDJTTIPXOJO'JHVSFXJUI9HJWFOPOUIF
IPSJ[POUBMBYJT*UXPVMEBQQFBSUIBUUIFBDDFQUBODF
P(H&) P( H) SBOHFGPSUIJTTUBUJTUJDPGoUPPVHIUUPCF
P( D) P(D&)
DPOTJEFSFE5IFBDUVBMSJTLPGGBJMJOHUIFUFTUCZ
SBOEPNDIBODFDBOCFDBMDVMBUFEBTGPMMPXT
Test
result 1SPCBCJMJUZUIBUU< =
5%*45 "#4
+
=
P(H&) P( H)

P(H&) 1SPCBCJMJUZUIBUU>= 15%*45 +

P( D)
=
4PUIFUPUBMQSPCBCJMJUZPGGBJMJOHUIFUFTUCZ
P(D) P(H) SBOEPNDIBODFXJMMCF PSVOEFS
D H
5IFBCPWFUUFTUJTDBMMFEBUFTUPGFRVBMJUZ'PS
Disease state
TVDIBUFTU UIFPOMZDPODFSOTBSFBCPVUUIF5ZQF
*FSSPSPSGBMTFBMBSNSBUFXIFOEFTJHOJOHUIFUFTU
)PXFWFSUIFQSPCBCJMJUZPGGBJMJOHUPEFUFDUBO
DSJUFSJBXBTTFUGPSUIJTTUBUJTUJDUIBUHBWFBMPXSJTL JNQPSUBOUEJGGFSFODF BMTPDBMMFEB5ZQF**FSSPS
PGGBJMVSFCZSBOEPNWBSJBUJPOBMPOF)PXFWFS XF PSCSPLFOBMBSNSBUF
JTBDPOTJEFSBUJPOXIFOXF
EPOPULOPXUIFEJTUSJCVUJPOPGFJUIFSUIFNFBOT BSFUSZJOHUPHBUIFSFWJEFODFPGFRVJWBMFODF#PUI
PSTUBOEBSEEFWJBUJPOT)PXEPXFQSPDFFE " FRVBMJUZBOEFRVJWBMFODFUFTUTXJMMCFEJTDVTTFEJO
TPMVUJPOUPUIJTQSPCMFNXBTOPULOPXOCFGPSF NPSFEFUBJMJOGVUVSFJOTUBMMNFOUTPGi4UBUJTUJDBM
UIFUIDFOUVSZ5IFOJO8JMMJBN(PTTFUU B 7JFXQPJOUw
CSFXFSXPSLJOHBUUIF(VJOOFTTCSFXFSZJO%VCMJO
QVCMJTIFEUIFBQQSPQSJBUFTUBUJTUJDVOEFSUIF ASSESSING THE PREDICTIVE VALUE
QTFVEPOZNi4UVEFOUw
)JTJOHFOJPVTTPMVUJPO OF A SCREENING TEST USING THE
TVHHFTUTUIBUXFGPSNUIFGPMMPXJOHi4UVEFOUTUw BERNOULLI DISTRIBUTION AND BAYES
statistic as a decision criterion. RULE
"EJBHOPTUJDTDSFFOJOHUFTUGPSBEJTFBTFTVDI
BTDBODFSPSIFQBUJUJTJTNFBOUUPCFBGBTU TBGF
BOESFMBUJWFMZJOFYQFOTJWFQSPDFEVSF4VDIUFTUT
Xtest Xcontrol PGUFOIBWFBOBQQSFDJBCMFGBMTFQPTJUJWFSBUF
t $POTFRVFOUMZBQPTJUJWFSFTVMUJTVTVBMMZGPMMPXFE
(Ntest 1)S (Ncontrol 1)S2control
( (
2
test 1 1 VQXJUIBNPSFJOWBTJWFBOEDPTUMZ CVUNPSF

Ntest Ncontrol 2 Ntest Ncontrol EJTDSJNJOBUJOHDPOGJSNBUPSZUFTU)PXFWFS
SFDFJWJOHBQPTJUJWFSFTVMUPOTVDIBTDSFFOJOHUFTU
DBOCFBTUSFTTGVMFYQFSJFODFGPSBQBUJFOUVOUJM
XIFSF JOPVSDBTF /test = Ncontrol5IF UIFNPSFEJTDSJNJOBUJOHDPOGJSNBUPSZUFTUJTBCMF
OVNFSBUPSPGUIJTTUBUJTUJDJTTJNQMZUIFFTUJNBUFE (hopefully) to discount the original screening
NFBOEJGGFSFODFBOEUIFEFOPNJOBUPSJTUIF result.
FTUJNBUFEQPPMFETUBOEBSEEFWJBUJPOPGUIFNFBO +VTUIPXNVDIOFXJOGPSNBUJPOEPFTBTDSFFOJOH
EJGGFSFODF5IFUSJDLPGOPSNBMJ[JOHBTUBUJTUJD UFTUQSPWJEF )PXEPFTSFDFJWJOHBQPTJUJWF
CZJUTFTUJNBUFETUBOEBSEEFWJBUJPOJTDBMMFE TDSFFOJOHSFTVMUDIBOHFUIFQSPCBCJMJUZUIBUB
i4UVEFOUJ[JOHw5IFSFTVMUJOHUTUBUJTUJDGPMMPXTB QBUJFOUIBTUIFEJTFBTF *TUIFTDSFFOJOHUFTUDPTU
DPOUJOVPVT4UVEFOUTUEJTUSJCVUJPO TFF5BCMF**
FGGFDUJWF 4VDIRVFTUJPOTBSFOPUPOMZJNQPSUBOU
with 1 = Ntest + Ncontrol EFHSFFTPGGSFFEPN UPQBUJFOUTBOEEPDUPST CVUBMTPUPNBOVGBDUVSFST
DPODFSOFEXJUIUIFRVBMJUZBOEBQQSPQSJBUFOFTT
5IFDVNVMBUJWFEJTUSJCVUJPOGVODUJPOPGUDBOCF PGTVDIQSPEVDUT BOEUPUIFSFHVMBUPSTXIPBTTVSF
PCUBJOFEBTGPMMPXT UIFTBGFUZBOEFGGFDUJWFOFTTPGUIFTFEFWJDFTGPSUIF
QVCMJD
1SPCBCJMJUZUIBUUXJMMCFMFTTUIBO9= *OUIFXPSLPG3FWFSFOE5IPNBT#BZFT
5%*45 "#4 9
+
GPS9 BOE BIVNCMF&OHMJTI1SFTCZUFSJBONJOJTUFS XBT

David LeBlond
QVCMJTIFEQPTUIVNPVTMZCZIJTGSJFOE3JDIBSE1SJDF
Figure 8: Population as the area of a unit square, subgroups.

)JTNFUIPEJTXPSUIDPOTJEFSJOHCFDBVTFJU
JMMVTUSBUFTUIFWBMVFPGJODMVEJOHQSJPSLOPXMFEHF
FH EJTFBTFQSFWBMFODFJOBQPQVMBUJPOCFJOH
TDSFFOFE
XIFOJOUFSQSFUJOHOFXEBUB FH SFTVMUT
P(D&) P(D )
of a recent screening test). P(D ) P(D&)
"TDSFFOJOHUFTUHFOFSBMMZIBTPOMZUXPPVUDPNFT D
PS5IFTUBUFPGIFBMUIPGBQFSTPOIBTPOMZUXP
MFWFMT)PS%5IFTFBSFBOBMPHPVTUPBDBUFHPSJDBM Disease
state
CJOPNJBMWBSJBCMFXJUIBTBNQMFTJ[FPG TFF5BCMF
*
5IJTTQFDJBMDBTFPGUIFCJOPNJBMEJTUSJCVUJPO P(H&) P(H )
JTSFGFSSFEUPBTUIF#FSOPVMMJEJTUSJCVUJPOBGUFS H
UIFUIDFOUVSZ*UBMJBONBUIFNBUJDJBO+BDPC P(H&)
P(H )
#FSOPVMMJ8PSLJOHXJUIUIF#FSOPVMMJEJTUSJCVUJPO
JTEFDFQUJWFMZTJNQMFCFDBVTFUIFQSPCBCJMJUZNBTT
P() P()
PGBSFTVMUJTFRVBMUPUIFQPQVMBUJPOQSPQPSUJPO
TFFUIF#FSOPVMMJiTUPSZwJO5BCMF*
"TQFDJBM
&YDFMGVODUJPOJTOPUOFFEFE5IFQSPQPSUJPOBTB Test result
QSPCBCJMJUZDBOCFXPSLFEXJUIEJSFDUMZ
'PSFYBNQMF +BNFTJTBQBUJFOUXIPNBZIBWF each depending on the proportion who would
BEJTFBTF+BNFTIBTOPUZFUCFFOTDSFFOFEGPSUIF SFDFJWFB VQQFS
PSo MPXFS
TDSFFOJOHUFTUSFTVMU
EJTFBTF8IBUJTUIFQSPCBCJMJUZUIBUIFIBTUIF 5IJTHFOFSBUFTGPVSUPUBMTVCTVCQPQVMBUJPOTXIPTF
EJTFBTF +BNFTNJHIUEPBO*OUFSOFUTFBSDIBOE KPJOUQSPQPSUJPOTJOUIFPSJHJOBMVOJUTRVBSFBSF
MFBSOUIFQSFWBMFODFPGUIFEJTFBTFJOQFPQMFMJLF 1 %
1 %
1 )
BOE1 )
5IFTZNCPM
IJNTFMG1SFWBMFODFJTEFGJOFEBTUIFQSPCBCJMJUZPG SFBETiBOEw*UJTDPNNVUBUJWFTPUIBU1 %

IBWJOHUIFEJTFBTF 1 %
GPSNFNCFSTPGBXFMM 1 %
"MPOHUIFMFGUBOESJHIUWFSUJDBMTJEFTPGUIF
EFGJOFEQPQVMBUJPO-FUTQSFUFOEUIBUUIJTFOUJSF VOJUTRVBSFBSFGPVSOFXWFSUJDBMRVBOUJUJFTXIPTF
QPQVMBUJPOIBECFFOTDSFFOFEBOEUIBUUIFUSVF MFOHUITDPSSFTQPOEUPUIFDPOEJUJPOBMQSPCBCJMJUJFT
EJTFBTFTUBUFJTLOPXO3FQSFTFOUJOHUIFQPQVMBUJPO 1 ]%
1 ]%
1 ])
BOE1 ])
5IFGJSTUBOE
BTUIFBSFBPGBVOJUTRVBSF BTTIPXOJO'JHVSF JT TFDPOEQSPCBCJMJUJFTSFGFSUPQSPQPSUJPOTPGUIF
VTFGVMCFDBVTFQSPCBCJMJUZDBOCFNBOJQVMBUFEBT EJTFBTFETVCQPQVMBUJPOBOEUIFUIJSEBOEGPVSUI
an area or a length. The area of any region within QSPCBCJMJUJFTSFGFSUPQSPQPSUJPOTPGUIFIFBMUIZ
UIFTRVBSFJTUBLFOBTBNFBTVSFPGUIFGSBDUJPOPG TVCQPQVMBUJPO
UIFQPQVMBUJPOUIBUNFFUDFSUBJODSJUFSJB5IJTMFBET 5IJOLPG1 %
BTUIFQSPQPSUJPOPGUIFBSFB
UPBHFPNFUSJDBMTPMVUJPOUP+BNFTRVFTUJPO'SPN PGUIFVOJUTRVBSFFRVBMUPUIFQSPQPSUJPOPG
UIFCBTFPGUIFVOJUTRVBSFXFEJWJEFUIFQPQVMBUJPO UIFQPQVMBUJPOUIBUJTCPUIEJTFBTFEBOEIBTB
XJUIBEBTIFEWFSUJDBMMJOFTFQBSBUJOHUIFEJTFBTFE screening result. Inspection of the upper left area
TVCQPQVMBUJPO POUIFMFGU
GSPNUIFIFBMUIZ 'JHVSFTIPXTUIBU
TVCQPQVMBUJPO POUIFSJHIU
5IFQSPCBCJMJUZUIBU
+BNFTJTIFBMUIZJTUIFO 1 %
1 ]%
1 %
<&RVBUJPO>
1 )
1 %
<&RVBUJPO> 4JNJMBSMZ UIFUPUBMQSPQPSUJPOPGUFTUSFTVMUTJO
UIFXIPMFVOJUTRVBSFQPQVMBUJPOJTTFFOBT
/PX+BNFTSFDFJWFTIJTTDSFFOJOHSFTVMU
6OGPSUVOBUFMZUIFSFTVMUJTQPTJUJWF
/PX XIBU 1
1 ]%
1 %
1 ])
1 )
<&RVBUJPO>
JTUIFQSPCBCJMJUZUIBUIFIBTUIFEJTFBTF HJWFO $POTJEFSJOH1 ])
BOE1 ])
BTQSPQPSUJPOT
UIJTQPTJUJWFSFTVMU 8FEFOPUFUIJTQSPCBCJMJUZ PGBVOJUMJOFUIBUBEEUP XFTFFGSPNUIFSJHIU
1 %]
XIFSFUIFWFSUJDBMCBSSFBETiDPOEJUJPOBM WFSUJDBMTJEFPG'JHVSFUIBU
POwPSiHJWFOUIBUw8IBUUIJTNFBOTJTUIBUXFBSF
SFTUSJDUJOH+BNFTUPBTVCQPQVMBUJPOXIPXPVME 1 ])
o1 ])
<&RVBUJPO>
IBWFBTDSFFOJOHSFTVMU+BNFTXBOUTUPLOPX
1 %]
BOEUPDPNQBSFJUUPUIFQSJPSQSPCBCJMJUZ 5IFTBNFQPQVMBUJPONBZBMTPCFTVCEJWJEFE
1 %
EJGGFSFOUMZ BTTIPXOJO'JHVSF JOUPGPVS
5PGJOE1 %]
XFNVTUGVSUIFSEJWJEFUIF TVCHSPVQTIBWJOHUIFTBNFQSPQPSUJPOTBTUIPTF
EJTFBTFE %
BOEIFBMUIZ )
TVCQPQVMBUJPOTJOUIF JO'JHVSF CVUXJUIEJGGFSFOURVBOUJUJFTBMPOHUIF
VOJUTRVBSFPG'JHVSFJOUPUXPTVCTVCQPQVMBUJPOT IPSJ[POUBMBOEWFSUJDBMBYFTPGUIFVOJUTRVBSF*O

David LeBlond
'JHVSFXFGJSTUEJWJEFUIFQPQVMBUJPOBDDPSEJOH GPMEJODSFBTF4PUIFOFXEBUBEPFTQSPWJEF
UPTDSFFOJOHUFTUSFTVMU POMFGU POSJHIU
UIFO TJHOJGJDBOUJOGPSNBUJPO)BE+BNFTSFDFJWFEBo
BDDPSEJOHUPEJTFBTFTUBUF %BCPWFBOE)CFMPX
TDSFFOJOHUFTUSFTVMU XFDPVMETJNJMBSMZPCUBJO
"OFYBNJOBUJPOPG'JHVSFBSFBTXJMMTIPXUIBU IJT1 )]
UIFQSFEJDUJWFWBMVFPGBOFHBUJWF
SFTVMU
0UIFSBQQMJDBUJPOTPG#BZFT3VMFXJMM
1 %]
1 %
1
<&RVBUJPO> CFBEESFTTFEJOGVUVSFDPMVNOTPGi4UBUJTUJDBM
7JFXQPJOUw
/PUJDFGSPNFRVBUJPOUIBUXFBSFNBLJOH
QSPHSFTTJOPCUBJOJOHUIFRVBOUJUZ1 %]
GPS CONCLUSIONS
+BNFT8JUIPVUCPSJOHUIFSFBEFSXJUIUIFBMHFCSB %FDJTJPONBLJOHJTPGUFOGBDJMJUBUFECZ
XFTJNQMZTUBUFUIBUFRVBUJPOTUISPVHINBZ USBOTGPSNJOHRVBMJUBUJWFEFTDSJQUJPOTBOE
CFDPNCJOFEUPPCUBJOUIFGPMMPXJOHTPMVUJPOGPS EJTDVTTJPOTJOUPRVBOUJUBUJWFQSPCBCJMJUZNPEFMT
1 %]
5IFQSFWJPVTFJHIUFYBNQMFTXFSFHJWFOUPJMMVTUSBUF
BGSBNFXPSLGPSVTJOHQSPCBCJMJUZDBMDVMBUJPOTUP
NBLFRVBOUJUBUJWFQSFEJDUJPOT QMBOSFTPVSDFT BOE
BTTFTTSJTLGSPNEBUBBOEPSQSJPSJOGPSNBUJPOGPS
P( D)P(D) TPNFDPNNPOQIBSNBDFVUJDBMEFDJTJPONBLJOH
P(D )
P( D)P(D)(1P( H))(1P(D)) TJUVBUJPOT5IJTBQQSPBDINBZCFTVNNBSJ[FEBT
follows:
t, OPXZPVSPCKFDUJWF*GUIJTJOWPMWFT
<&RVBUJPO> QSFEJDUJOHUIFWBMVFTPGSBOEPNWBSJBCMFT
WBSJBCMFEBUBTUBUJTUJDT PSVODFSUBJO
where QBSBNFUFST
UIFOVTJOHQSPCBCJMJUZ
EJTUSJCVUJPONFUIPETXJMMCFVTFGVM
]%
TFOTJUJWJUZPGUIFTDSFFOJOHUFTU
1 t6OEFSTUBOEUIFTBNQMJOHEJTUSJCVUJPOPGUIF
QSPCBCJMJUZPGBHJWFOEJTFBTF
EBUBTUBUJTUJDTZPVBSFDPMMFDUJOH*EFOUJGZUIF
type of data (discrete or continuous) and the
])
TQFDJGJDJUZPGUIFTDSFFOJOHUFTU
1 EJTUSJCVUJPOTUPSZBOEBTTVNQUJPOTUIBUBQQMZ
QSPCBCJMJUZPGBoHJWFOIFBMUI
to your situation.
t5SZ JGQPTTJCMF
UPGJOEBNBUDICFUXFFOZPVS
&RVBUJPO DBMMFEi#BZFT3VMFwJOIPOPSPG PXOEBUBHFOFSBUJPOiTUPSZwBOEUIBUPGB
5IPNBT#BZFT JTGJOEJOHNPSFBQQMJDBUJPOTJO DPNNPOEJTUSJCVUJPO TVDIBTUIPTFJO5BCMFT*
UPEBZTFSBPGGBTUDPNQVUFST1 %
JTSFGFSSFE or II.
UPBTUIFQSJPSQSPCBCJMJUZ PGEJTFBTF
CFGPSF t#FDPNFGBNJMJBSXJUIQSPCBCJMJUZGVODUJPOTJO
EBUBJTBWBJMBCMF BOE1 %]
JTSFGFSSFEUPBTUIF SFBEJMZBWBJMBCMFTPGUXBSFTVDIBT.JDSPTPGU
QPTUFSJPSQSPCBCJMJUZ PGEJTFBTF
HJWFOUIBUEBUBJT &YDFM4PNFPGUIFTFBSFQSFTFOUFEJO5BCMFT*
BWBJMBCMF8FDBOOPXVTF#BZFT3VMFUPBOTXFS and II.
t$ POTJEFSXIFUIFSJUJTBQQSPQSJBUFUPJODMVEF
+BNFTRVFTUJPO-FUTUBLFUIFGPMMPXJOHBTJOQVUUP
QSJPSJOGPSNBUJPO FH FYQFSUPQJOJPO
FRVBUJPO
QSJPSSFMFWBOUEBUB UIFPSZ
BTQBSUPGZPVS
QSFEJDUJPOT*GTP USZUPSFEVDFUIBUTVCKFDUJWF
%
UIBUJTPGUIFQPQVMBUJPOIBTUIF
1
QSJPSJOGPSNBUJPOUPQSPCBCJMJUZEJTUSJCVUJPOT
disease
XJUIBQQSPQSJBUFTIBQFBOETQSFBE"#BZFTJBO
1 ]%
UFTUTFOTJUJWJUZ
BQQSPBDINBZQSPWFGSVJUGVM
1 ])
UFTUTQFDJGJDJUZ
t%POPUGBJMUPDPOTVMUBUSBJOFETUBUJTUJDJBO
.PTUQFPQMFXPVMEDPOTJEFSBTDSFFOJOHUFTU
JODBTFTXIFSFUIFPCKFDUJWFT BQQSPQSJBUF
XJUIBTFOTJUJWJUZBOETQFDJGJDJUZB
EJTUSJCVUJPO BTTVNQUJPOT PSUIFSJHIUBQQSPBDI
IJHIMZEJTDSJNJOBUJOHUFTU)PXFWFSXFDBO
appear unclear.
TIPXGSPN#BZFT3VMFUIBUXIFOBQQMJFEUP
BQPQVMBUJPOJOXIJDIPOMZPGUIFQFPQMF 4UBUJTUJDTBOEQSPCBCJMJUZBSFNBUVSFEJTDJQMJOFT
IBWFUIFEJTFBTF UIFQSFEJDUJWFWBMVFPGB XJUIBUMFBTUBZFBSEFWFMPQNFOUIJTUPSZ
SFTVMU 1 %]
JTPOMZ5IVTUIFQSPCBCJMJUZ 8IBUFWFSZPVSQBSUJDVMBSSJTLBTTFTTNFOUPS
UIBU+BNFTIBTUIFEJTFBTFJTMFTTUIBOIBMGJU QSFEJDUJPOQSPCMFN JUJTIJHIMZMJLFMZUIBUBHPPE
JTNPSFMJLFMZUIBUIFJTIFBMUIZUIBOEJTFBTFE BQQSPBDIUPNFFUJOHZPVSPCKFDUJWFJTXFMMLOPXO
/PUJDFIPXFWFSUIBUUIJTOFXEBUB BTDSFFOJOH *OUIFJOUFSFTUPGQSPNPUJOHCFTUQSBDUJDFT BXJEF
SFTVMU
JODSFBTFEUIFQSPCBCJMJUZUIBU+BNFTIBT SBOHFPGTVDIBQQSPBDIFTXJMMCFQSFTFOUFEJOGVUVSF
UIFEJTFBTFGSPN UIFQSFWBMFODF
UP B JOTUBMMNFOUTPGi4UBUJTUJDBM7JFXQPJOUw

David LeBlond
measurement) to some critical failure

GLOSSARY
range. In a one-sided test, the test can fail
Bayes rule: A process for combining the information
on only one side (i.e., statistic > an upper limit
in a data set with relevant prior information
or statistic < a lower limit, but not both). In a
(theory, past data, expert opinion, and
two-sided test, both an upper and lower limit
knowledge) to obtain posterior information.
are used and thus the test can fail on either
Prior and posterior information are
side.
expressed in the form of prior and posterior
probability distributions, respectively, of the Pooled variance: An estimate of a population variance
underlying physical parameters. obtained from two or more separate samples
obtained from the same population or at
Bernoulli distribution: A special case of the Binomial
least from populations having the same
distribution in which the sample size is equal to
population variance. The square root of the
1. A variable following the Bernoulli distribution
pooled variance is called the pooled standard
has only two possible levels: 0 or 1. The
deviation.
probability mass associated with the 1 level is
equal to the proportion parameter, . Posterior probability: An improved estimate of a
proportion that is obtained by using Bayes
Broken alarm rate: The ability of a statistical test to
rule to combine information from a subjective
detect a difference that is large enough to
prior probability with that available from
be practically important (e.g., the difference
experimental data.
between two means being greater than
is acceptable) is a consideration when Predictive value of a failing test result: The
gathering evidence for equivalence of two conditional probability that an object
processes. The probability that a given test under test (e.g., a person) is defective (e.g.,
will fail to detect such a difference is called diseased), given that the object failed the
the broken alarm rate, the false negative test (e.g., a + result on a diagnostic test).
rate, a Type II error rate, or the beta error Prior probability: An estimate of a proportion
rate. that is based on theory, past data, expert
Conditional probability: The probability of an event (say opinion, and knowledge (e.g., prevalence
A) occurring, given that some other relevant of disease in a population, false positive or
event (say B) has occurred, symbolized P(A|B). negative rates of a screening test). Often the
For example, let A = house is on fire and B = proportion is a parameter of a probability
smoke alarm is sounding. distribution such as the binomial or Bernoulli
distribution.
False alarm rate: Because all statistical tests (e.g.,
control chart rules, comparison with Predictive value of a passing test result: The
acceptance limits, t- or F-tests ) are based conditional probability that an object under
on a comparison of a random variable (e.g., test (e.g., a person) is acceptable (e.g.,
some categorical or continuous measure healthy), given that the object passed the test
calculated from data) to some fixed (e.g., a - result on a diagnostic test).
decision limit, there is generally a non-zero Sensitivity of a test: The conditional probability of failing
probability of exceeding the limit (i.e., a test (e.g., a + result on a diagnostic test),
failing the test) when, in fact, the process given that the object under test (e.g., a person)
is performing acceptably. This probability is is truly defective (e.g., diseased).
called the false alarm rate (also referred to Skewness: One characteristic of a continuous
as the false positive, the Type I or alpha probability distribution that indicates the
error rate). degree of symmetry of the distribution about
Multiplicity: When multiple statistical tests (e.g., control its center (i.e., mean, mode, or median).
chart rules) are applied to different aspects (e.g., Distributions showing longer tails at lower
trending patterns) of a data set, the overall false or higher values are said to be left or right
alarm rate of any one test failing may be greater skewed, respectively. The normal distribution
than that of any single test when applied alone. has no skewness and is perfectly symmetrical
This statistical phenomenon is referred to as about its mean. The lognormal distribution is
multiplicity. skewed to the right.
One-sided or two-sided test: Statistical tests often Specificity of a test: The conditional probability of
involve the comparison of some random passing a test (e.g., a - result on a diagnostic
statistic, calculated from data (such as test), given that the object under test (e.g., a
an F-ratio, a t value, a count, or other person) is truly acceptable (e.g., healthy).

David LeBlond
Statistic: A numeric value calculated from data. Often

8 JMMJBN 5IPNQTPO
i&MFD U S JD B M 6OJU T PG
a statistic serves as an estimate of some
.FBTVSFNFOU wPopular Lectures and Addresses,7PMVNF
underlying parameter of an appropriate
Western Electric Statistical Quality Control Handbook 8FTUFSO
probability distribution (e.g., the sample
&MFDUSJD$PSQPSBUJPO *OEJBOBQPMJT *OEJBOB
average and standard deviation are estimates
.POUHPNFSZ % Introduction to Statistical Quality Control
of the and parameters, respectively, of the
OEFEJUJPO +PIO8JMFZ4POT /FX:PSL QBHF
normal probability distribution from which the
i4UVEFOU wi5IFQSPCBCMFFSSPSPGBNFBO wBiometrika
measurements are drawn).
QBHF

Subpopulation: a subset of a population. 1SJDF 3
i"OFTTBZUPXBSETTPMWJOHBQSPCMFNJOUIF
Sub-Subpopulation: A subset of a subpopulation. EPDUSJOFPGDIBODFT w%BMF A Most Honourable Remembrance:
The life and Work of Thomas Bayes 4QSJOHFS /FX:PSL
REFERENCES

-F#MPOE % i%BUB 7BSJBUJPO 6ODFSUBJOUZ BOE1SPCBCJMJUZ 3PTOFS # Fundamentals of Biostatistics 3rd edition %VYCVSZ
%JTUSJCVUJPOT wJournal of GxP Compliance 7PM /P QQ 1SFTT #FMNPOU $BMJGPSOJB QQ JVT
4QSJOH

Paul L. Pluta
Case Study #3Process Validation

Failure of a Liquid Product
Batch Size IncreaseIdentical
Manufacturing Tanks
Paul L. Pluta
Validation Case Studies discusses situations useful VDUTPMVUJPO5IFWBMJEBUJPOQFSGPSNBODF

to practitioners in validation. Each case presented RVBMJGJDBUJPO 12
XBTBGBJMVSF.JYJOHJOUIF
deals with a specific validation problem, elements of UBOLXBTOPUTVGGJDJFOUMZSJHPSPVTUPEJTTPMWF
which are described to demonstrate strategy to solve UIFGPSNVMBUJPOJOHSFEJFOUT
the problem. We intend this column to be a useful t5IFTVQQPTFEMZJEFOUJDBMUBOLEJEOPUIBWFUIF
resource for daily work applications. The main objec- TBNFJNQFMMFSBSSBOHFNFOUBTUIFPUIFSMBSHF
tive of this column: Useful and practical information. NJYJOHUBOLTBUUIFTJUF"MUIPVHIUIFWPMVNF
Reader comments, questions, and suggestions are PGBMMUBOLTXFSFUIFTBNF UIFNJYJOHJNQFMMFST
needed to help us fulfill our objective for this column. were different.
Case studies illustrating validation issues submitted t"OFXNBOVGBDUVSJOHQSPDFTTXBTEFWFMPQFE
by readers are most welcome. Please send your com- VTJOHBSFEVDFETPMVUJPOWPMVNFBUBOJOUFS
ments and suggestions to journal coordinating editor NFEJBUFQSPDFTTTUBHFUIBUFOBCMFETVDDFTTGVM
Susan Haigney at shaigney@advanstar.com. NBOVGBDUVSJOH
t$PSSFDUJWFBDUJPOBOEQSFWFOUJWFBDUJPO
KEY POINTS DISCUSSED $"1"
BDUJWJUJFTJODMVEFEBOFXNBOVGBDUVS
The following key points are discussed: JOHQSPDFTT USBJOJOH EPDVNFOUBUJPO QSPDFTT
t"DBTFTUVEZJOWPMWJOHQSPDFTTWBMJEBUJPOPGB WBMJEBUJPO QPTUWBMJEBUJPONPOJUPSJOH FRVJQ
CBUDITJ[FJODSFBTFGPSBMJRVJETPMVUJPOQSPE NFOURVBMJGJDBUJPO BOEBQQMJDBUJPOPGMFTTPOT
VDUJTEFTDSJCFE learned to other site products.
t5IFCBUDITJ[FJODSFBTFXBTDPOTJEFSFEUPCFB t&RVJQNFOURVBMJGJDBUJPOBOEFRVJQNFOU
SFMBUJWFMZTJNQMFDIBOHFCFDBVTFPUIFSTJNJMBS FRVJWBMFODFEPDVNFOUBUJPOXFSFSFWJFXFE
QSPEVDUTBUUIFTJUFXFSFBMSFBEZNBOVGBD BOENPEJGJFE
UVSFEBUUIFTBNFJODSFBTFECBUDITJ[F BOEUIF t4FWFSBMPUIFSQSPEVDUTXFSFBMTPNBOVGBDUVSFE
FRVJQNFOUUPCFVTFEXBTUIPVHIUUPCFJEFO VTJOHUIFOFXNBOVGBDUVSJOHQSPDFEVSF
UJDBMUPPUIFSTJUFFRVJQNFOU t.JOPSFRVJQNFOUEJGGFSFODFTNBZIBWFBWFSZ
t.BOVGBDUVSJOHXBTTUBSUFEBOEUIFGPSNVMB TJHOJGJDBOUFGGFDUJONBOVGBDUVSJOHi*EFOUJDBMw
UJPOJOHSFEJFOUTEJEOPUEJTTPMWFJOUIFQSPE NBZSFBMMZOPUCFTP
[
ABOUT THE AUTHORS
For more Author Paul L. Pluta, Ph.D., is a pharmaceutical scientist with extensive technical and management expe-
information, rience in the pharmaceutical industry. Dr. Pluta is also adjunct associate professor at the University
go to of Illinois College of Pharmacy in Chicago. He edited and contributed chapters of Cleaning and
gxpandjvt.com/bios Cleaning Validation, published by PDA and Davis Healthcare International. He may be reached by e-
mail at paul.pluta@comcast.net.

Paul L. Pluta
INTRODUCTION BOEIZESPDIMPSJDBDJEPSTPEJVNIZESPYJEF5IF
5IJTDBTFTUVEZXBTQSPWJEFEUPUIFJournal of Valida- NBOVGBDUVSJOHQSPDFTTGPSUIFQSPEVDUDPNQSJTFE
tion TechnologyCZBSFBEFSXIPSFRVFTUFEBOPOZNJUZ the following steps:
5IFFWFOUEFTDSJCFEJTBOBDUVBMPDDVSSFODF t"QQSPYJNBUFMZPGQVSJGJFEXBUFSXBTBEEFE
"TNBMMNPMFDVMFQIBSNBDFVUJDBMDPNQBOZ UPUIFNJYJOHUBOL4PMVUJPOWPMVNFXBTNFB
SFRVJSFEBCBUDITJ[FJODSFBTFGPSBMJRVJETPMVUJPO sured with a dip stick.
QSPEVDUUPNFFUJODSFBTFEDPNNFSDJBMEFNBOET5IF t5IFNJYFSXBTTUBSUFE
CBUDITJ[FJODSFBTFXBTDPOTJEFSFEUPCFBSFMBUJWFMZ t"MMTPMJEJOHSFEJFOUTJODMVEJOHUIFBDUJWFESVH
TJNQMFDIBOHFCFDBVTFPUIFSTJNJMBSQSPEVDUTBUUIF XFSFBEEFEUPUIFNJYJOHUBOLUISPVHIBQPSUBU
TJUFXFSFBMSFBEZNBOVGBDUVSFEBUUIFTBNFJODSFBTFE UIFUPQPGUIFNJYJOHUBOL1PXEFSBEIFSJOHUP
CBUDITJ[F5IFOFFEGPSWBMJEBUJPOXBTRVFTUJPOFE the side of the tank was rinsed into the solution
CFDBVTFUIFQSPEVDUBOEQSPDFTTBQQFBSFEUPCF XJUIBTNBMMBNPVOUPGQVSJGJFEXBUFS
JEFOUJDBMUPPUIFSQSPEVDUNBOVGBDUVSJOH BOEUIF t"GUFSBMMJOHSFEJFOUTXFSFBEEFE UIFTPMVUJPO
FRVJQNFOUUPCFVTFEXBTUIPVHIUUPCFJEFOUJDBMUP XBTNJYFEGPSNJOVUFT
BMMPUIFSNJYJOHUBOLT"HSFFNFOUPOBPOFMPUDPO t4PMVUJPOQ)XBTNFBTVSFE*GOFDFTTBSZ Q)
GJSNBUPSZQFSGPSNBODFRVBMJGJDBUJPO 12
WBMJEBUJPO XBTBEKVTUFEXJUIIZESPDIMPSJDBDJETPMVUJPOPS
CBUDIXBTVMUJNBUFMZEFDJEFECFDBVTFUIFDIBOHFXBT TPEJVNIZESPYJEFTPMVUJPO
DPOTJEFSFEUPCFPGNJOJNBMSJTL"OJOGSFRVFOUMZ t"EEJUJPOBMQVSJGJFEXBUFSXBTBEEFEUPNBLF
VTFENJYJOHUBOLPGUIFBQQSPQSJBUFTJ[FUPNBLFUIF UIFGJOBMTPMVUJPOWPMVNF'JOBMWPMVNFXBT
MBSHFCBUDITPMVUJPOXPVMECFVTFE.BOVGBDUVSJOH NFBTVSFEXJUIBEJQTUJDL
XBTTUBSUFE5IFVOFYQFDUFEIBQQFOFENBOV t5IFTPMVUJPOXBTNJYFEGPSNJOVUFT
GBDUVSJOHIBEUPCFBCPSUFEBGUFSTFWFSBMIPVSTPG t4PMVUJPOXBTGJMUFSFEBOEGJMMFEJOUPUIFDPN
NJYJOH NVDIMPOHFSUIBOUIFQSPDFTTTIPVMEIBWF NFSDJBMQBDLBHJOH
SFRVJSFE*OHSFEJFOUTEJEOPUEJTTPMWFJOUIFQSPEVDU
TPMVUJPO5IFWBMJEBUJPOXBTBGBJMVSF VALIDATION EVENT
5IJTEJTDVTTJPOQSPWJEFTUIFGPMMPXJOH "TNBMMNPMFDVMFQIBSNBDFVUJDBMDPNQBOZSFRVJSFE
tProcess description background5IFNBO BCBUDITJ[FJODSFBTFGPSBOPSBMMJRVJETPMVUJPO
VGBDUVSJOHQSPDFTTGPSUIFJODSFBTFECBUDITJ[F QSPEVDU1MBOUNBOBHFNFOUBOETVQQMZDIBJO
QSPEVDUJTCSJFGMZEFTDSJCFE QFSTPOOFMPSEFSFEBOJODSFBTFECBUDITJ[FGPS
tCompliance event"EFTDSJQUJPOPGUIFFWFOU DPNNFSDJBMTVQQMZBOEDPTUSFEVDUJPOQVSQPTFT
UIFLFZJTTVFTUPCFBEESFTTFE BOEBQQMJDBCMF &RVJQNFOUUPBDDPNQMJTIUIFJODSFBTFECBUDITJ[F
DVSSFOUHPPENBOVGBDUVSJOHQSBDUJDF $(.1
XBTBWBJMBCMFBOJOGSFRVFOUMZVTFENJYJOHUBOL
SFRVJSFNFOUT PGUIFBQQSPQSJBUFTJ[FXPVMECFVTFEUPNBLFUIF
tInvestigation*OUFSWJFXTBOEBDUJPOTDPOEVDU MBSHFCBUDI5IFNJYJOHUBOLXBTUIFTBNFTJ[FBT
FEUPJOWFTUJHBUFUIFFWFOU TFWFSBMPUIFSUBOLTJOUIFGBDJMJUZBOEXBTUIPVHIU
tDiscussion,FZJOGPSNBUJPO BDUJWJUJFT BOE UPCFJEFOUJDBMUPBMMPUIFSUBOLT)PXFWFS JUXBT
analysis. BCBDLVQUBOLUIBUIBEOPUCFFOVTFEGPSTFWFSBM
tCorrective action and preventive action ZFBST5IFWBMJEBUJPOHSPVQQSFQBSFEBQSPDFTT
(CAPA)"DUJPOTBOEJNQSPWFNFOUTJNQMF WBMJEBUJPOQSPUPDPMBTSFRVFTUFECZQMBOUNBOBHF
NFOUFE5IFTFJODMVEFEEFWFMPQNFOUPGBOFX NFOU5IFQMBOUNBOVGBDUVSFETFWFSBMPUIFSPSBM
NBOVGBDUVSJOHQSPDFTT USBJOJOH EPDVNFOUB TPMVUJPOQSPEVDUTBUUIFTBNFJODSFBTFECBUDITJ[F
UJPO QSPDFTTWBMJEBUJPO QPTUWBMJEBUJPONPOJUPS JOPUIFSUBOLT.BOBHFNFOUBTTVNFEUIBUUIFSF
JOH FRVJQNFOURVBMJGJDBUJPO BOEBQQMJDBUJPOUP XPVMECFOPQSPCMFNXJUIUIFJODSFBTFECBUDI
PUIFSTJNJMBSQSPEVDUTBUUIFTJUF TJ[FBOEPSEFSFEBOJNNFEJBUFCBUDITJ[FJODSFBTF
tPost CAPA .BJOUBJOJOHWBMJEBUJPOBOEQFSGPS 5IFQSPEVDUIBEBHPPENBOVGBDUVSJOHBOERVBM
NBODF0UIFSBDUJPOTBOEJNQSPWFNFOUTJNQMF JUZIJTUPSZBUUIFTNBMMFSCBUDITJ[F0OFMPUXBT
NFOUFEBTBSFTVMUPGLOPXMFEHFHBJOFEBOE SFRVFTUFEUPWBMJEBUFUIFJODSFBTFECBUDITJ[F
BDUJWJUJFTBTTPDJBUFEXJUIUIFJODJEFOU CFDBVTFTFWFSBMPUIFSTJNJMBSQSPEVDUTXFSFBMSFBEZ
NBOVGBDUVSFEBUUIJTTBNFMBSHFCBUDITJ[F5IF
PROCESS DESCRIPTION BACKGROUND FYJTUJOHCBUDISFDPSEEPDVNFOUBUJPOXBTNPEJGJFE
5IFNBOVGBDUVSJOHQSPDFTTJOUIJTDBTFTUVEZ UPNBOVGBDUVSFUIFJODSFBTFECBUDITJ[F/PQSPC
JOWPMWFENBOVGBDUVSFPGBOBRVFPVTTPMVUJPOJOB MFNTXFSFFYQFDUFE
TUBOEBSENJYJOHUBOLXJUIDFOUSBMJNQFMMFS*OHSF .BOVGBDUVSJOHPQFSBUPSTNBOVGBDUVSFEUIFMBSHFS
EJFOUTJOUIFGPSNVMBUJPOJODMVEFEUIFBDUJWFESVH CBUDITJ[FBDDPSEJOHUPUIFTQFDJGJFEEJSFDUJPOT5IF
IZESPQIJMJDQPMZNFSUPJODSFBTFWJTDPTJUZ CVGGFS QSPDFEVSFVTFEXBTTDBMFEGSPNUIFTNBMMFSCBUDI
TBMUTUPNBJOUBJOQ) QSFTFSWBUJWFT GMBWPS DPMPS TJ[FQSPDFEVSF5IFQSPDFEVSFVTFEXBTBMTPUIF

Paul L. Pluta
TBNFBTVTFEGPSPUIFSQSPEVDUTBUBTJNJMBSCBUDI Figure: Mixing tank with different impellers.

TJ[F8IFODFSUBJOJOBDUJWFJOHSFEJFOUTXFSFBEEFE
UPUIFNJYJOHUBOL UIFTFNBUFSJBMTGMPBUFEPOUIF
UPQPGUIFXBUFSJOUIFUBOLCVUEJEOPUEJTTPMWF
.JYJOHDPOUJOVFEGPSTFWFSBMIPVST5IFJOBDUJWF
JOHSFEJFOUTTUJMMEJEOPUEJTTPMWF5IFNBOVGBDUVS
JOHQSPDFTTXBTUFSNJOBUFE5IFWBMJEBUJPO12DPO
GPSNBODFMPUIBEGBJMFE
What are the Issues?

5IFSFXFSFTFWFSBMDSJUJDBMJTTVFTUPCFJOWFTUJHBUFE
as follows:
t%JENBOVGBDUVSJOHPQFSBUPSTGPMMPXCBUDI
SFDPSEEJSFDUJPOTQSPQFSMZ
t%JEBMMHSPVQTBQQSPWFUIFOFXCBUDISFDPSE
t%JEUIFUFDIOJDBMHSPVQSFWJFXUIFQSPDFTTBOE
BQQSPWFUIFOFXCBUDISFDPSE
t8IZEJEUIFQSPDFTTWBMJEBUJPOGBJM JODSFBTFECBUDITJ[F5IFDIBOHFIBECFFOFWBMV
t8BTTPNFUIJOHXSPOHXJUIUIFFRVJQNFOU BUFECZUIFWBMJEBUJPOBQQSPWBMDPNNJUUFFBOEPOF
12DPOGPSNBODFMPUXBTSFDPNNFOEFE5IFSFXBT
CGMP Requirements OPFYQFDUBUJPOUIBUUIFQSPDFTTXPVMEOPUCFDPN
3FMFWBOUHPPENBOVGBDUVSJOHQSBDUJDF (.1
pletely successful. Other products in the plant had
SFRVJSFNFOUTBQQMJDBCMFUPUIFFWFOUBSFMJTUFEBT CFFOTVDDFTTGVMMZNBOVGBDUVSFEBUUIFTBNFCBUDI
follows: TJ[F5IFTFMPUTIBECFFONBOVGBDUVSFEJOXIBU
t4VCQBSU%&RVJQNFOU XBTUIPVHIUUPCFFRVJWBMFOUNJYJOHUBOLT7BMJEB
t&RVJQNFOU$MFBOJOHBOE.BJOUFOBODF UJPOEPDVNFOUBUJPOJOEJDBUFEUIBUBMMNJYJOHUBOLT
t4VCQBSU+3FDPSETBOE3FQPSUT PGUIJTWPMVNFXFSFFRVJWBMFOU
t(FOFSBM3FRVJSFNFOUT
t#BUDI1SPEVDUJPOBOE$POUSPM3FDPSET Technical Personnel Evaluation
t1SPEVDUJPO3FDPSE3FWJFX 5FDIOJDBMQFSTPOOFMBGGJSNFEUIBUUIFSFTIPVME
IBWFCFFOOPQSPCMFNXJUINBOVGBDUVSJOHUIF
INVESTIGATION JODSFBTFECBUDITJ[FVOEFSOPSNBMDJSDVNTUBODFT
*OWFTUJHBUJPOBOEVMUJNBUFSFTPMVUJPOPGUIJTFWFOU 5IFNBOVGBDUVSJOHQSPDFTTXBTTUSBJHIUGPSXBSEBT
SFRVJSFEJOWPMWFNFOUPGTFWFSBMHSPVQT5IFTF follows:
JODMVEFEQFSTPOOFMJOWPMWFEJOUIFJODJEFOU FH t"QQSPYJNBUFMZPGQVSJGJFEXBUFSXBTBEEFE
NBOVGBDUVSJOHBOERVBMJUZBTTVSBODF<2">
BOE UPUIFNJYJOHUBOL
UFDIOJDBMQFSTPOOFMSFTQPOTJCMFGPSUIFNBOVGBD t4PMJEJOHSFEJFOUTXFSFBEEFEUPUIFNJYJOHUBOL
UVSJOHQSPDFTTBOEQSPDFTTWBMJEBUJPO5IFSFXFSF t5IFTPMVUJPOXBTNJYFEGPSNJOVUFT
NBOZEFUBJMTUIBUOFFEFEUPCFJOWFTUJHBUFEPSDPO "MMJOHSFEJFOUTXFSFFYQFDUFEUPEJTTPMWFXJUIJO
GJSNFE1FSTPOOFMGSPNBMMHSPVQTXFSFJOUFSWJFXFE UIFNJOVUFNJYJOHUJNF
and interacted to address the issues.
Key Information
Manufacturing Personnel Interviews 5IFHFOFSBMNBOVGBDUVSJOHQSPDFTTXBTVTFEUP
.BOVGBDUVSJOHQFSTPOOFMBGGJSNFEUIBUBMMCBUDI NBOVGBDUVSFBTNBMMCBUDITJ[FNBOZUJNFT5IJT
SFDPSEEJSFDUJPOTXFSFQFSGPSNFEBTTQFDJGJFE TBNFQSPDFTTXBTBQQSPQSJBUFMZTDBMFEGPSVTFXJUI
.BOVGBDUVSJOHPQFSBUPSTIBENBOVGBDUVSFEFRVJWB UIFJODSFBTFECBUDITJ[F5IFFOHJOFFSJOHTUBGGXBT
MFOUQSPEVDUTBUMBSHFCBUDITJ[FVTJOHBOFRVJWBMFOU SFRVFTUFEUPFWBMVBUFUIFNJYJOHUBOLVTFEGPSUIF
QSPDFEVSFBUUIFTJUFNBOZUJNFT5IFZIBEBMTP GBJMFE12DPOGPSNBODFSVO5IFNJYJOHUBOLUIBU
NBOVGBDUVSFEUIFTQFDJGJDQSPEVDUPGUIJTDBTFTUVEZ XBTVTFEUPNBOVGBDUVSFUIFJODSFBTFECBUDITJ[F
BUTNBMMCBUDITJ[FNBOZUJNFT5IFSFXFSFOP XBTBOJOGSFRVFOUMZVTFEUBOL5IJTUBOLEJEOPU
USBJOJOHJTTVFTPSWBSJBUJPOJOQFSGPSNBODFBTTPDJ IBWFUIFTBNFJNQFMMFSBSSBOHFNFOUBTUIFPUIFS
BUFEXJUIUIFNBOVGBDUVSJOHQSPDFEVSF MBSHFNJYJOHUBOLTBUUIFTJUF"MMNJYJOHUBOLT
FYDFQUUIFUBOLVTFEGPSUIFQSPDFTTWBMJEBUJPO IBE
Batch Record Review and Approval UISFFQJUDIFECMBEFJNQFMMFSTMPDBUFEBUUIFUPQ
"MMSFMFWBOUQFSTPOOFMIBESFWJFXFEBOEBQQSPWFE NJEEMF BOECPUUPNPGUIFNJYFSTIBGU5IFUBOL
UIFOFXNBOVGBDUVSJOHCBUDISFDPSEXJUIUIF VTFEGPSUIFGBJMJOHCBUDIIBEPOMZUXPQJUDIFE

Paul L. Pluta
CMBEFJNQFMMFSTPOFMPDBUFEBUUIFCPUUPNBOEPOF EPDVNFOUBUJPOXBTBMTPEFWFMPQFE5ISFF12SVOT
MPDBUFENJEMFWFMPOUIFNJYFSTIBGU5IFNJYJOH XFSFSFDPNNFOEFECFDBVTFUIFNBOVGBDUVSJOHQSP
UBOLXJUIUXPJNQFMMFSTEJEOPUQSPWJEFBEFRVBUF DFTTXBTDPNQMFUFMZOFXUPUIFTJUF.BOVGBDUVS
BYJBMGMPXUPQVMMTPMJEJOHSFEJFOUTJOUPUIFCVML JOH 2" FOHJOFFSJOH BOEUFDIOJDBMTVQQPSUNBO
MJRVJEUPFGGFDUEJTTPMVUJPO5IFNJYJOHUBOLVTFEJO BHFNFOUBQQSPWFEUIFDIBOHFBOEBMMBTTPDJBUFE
UIFGBJMFE12SVOXBTTJHOJGJDBOUMZEJGGFSFOUGSPN EPDVNFOUBUJPO
BMMPUIFSNJYJOHUBOLTBUUIFTJUF)PXFWFS CFDBVTF
UIJTUBOLXBTBCBDLVQUBOLUIBUIBEOPUCFFOVTFE CORRECTIVE AND PREVENTIVE
GPSTFWFSBMZFBST OPPOFXBTBXBSFPGUIFEJGGFS ACTIONS
FODF"MUIPVHIUIFWPMVNFTPGBMMUBOLTXFSFUIF 5IFGPMMPXJOH$"1"BOEBTTPDJBUFEBDUJWJUJFTXFSF
TBNF UIFNJYJOHJNQFMMFSTXFSFWFSZEJGGFSFOU TFF conducted:
'JHVSF
/PPOFIBEMPPLFEJOTJEFUIFUBOL t/FXNBOVGBDUVSJOHQSPDFTT
tTraining
DISCUSSION t%PDVNFOUBUJPO
*OGPSNBUJPOPCUBJOFEUISPVHIJOUFSWJFXTBOE t1SPDFTTWBMJEBUJPO
TVCTFRVFOUFYQFSJNFOUBMXPSLFOBCMFEHPPE t1PTUWBMJEBUJPONPOJUPSJOH
VOEFSTUBOEJOHPGUIFQSPCMFNBOEVMUJNBUFTPMV t&RVJQNFOURVBMJGJDBUJPOBOEFRVJQNFOU
tion. Technical personnel conducted experi FRVJWBMFODF
NFOUBMSVOTBOEQSPQPTFEQSPDFTTJOHDIBOHFTUP t0UIFSTJNJMBSQSPEVDUTNBOVGBDUVSFEBUUIF
FOBCMFNBOVGBDUVSJOHJOBMMNJYJOHUBOLT5XP site.
FYQFSJNFOUBMSVOTXFSFDPOEVDUFE
New Manufacturing Process
Run Number One "OFXNBOVGBDUVSJOHQSPDFTTDPNQSJTJOHUIFGPM
5IFGJSTUSVODPNQSJTFEBEEJUJPOPGQVSJGJFEXBUFS MPXJOHXBTJNQMFNFOUFEXIFOUIFCBDLVQUBOLXBT
UPUIFNJYJOHUBOL5IFBNPVOUPGXBUFSBEEFEXBT used:
NFBTVSFEUPCFKVTUBCPWFUIFNJEEMFJNQFMMFS5IJT t"QQSPYJNBUFMZPGQVSJGJFEXBUFSXBTBEEFE
XBUFSMFWFMQSPWJEFEHPPENJYJOHBTFWJEFODFECZB UPUIFNJYJOHUBOL
TUSPOHWPSUFYBOEEJEOPUDBVTFFYDFTTJWFTQMBTIJOH t.JYJOHXBTTUBSUFE
"QQSPYJNBUFMZPGUIFGJOBMTPMVUJPOWPMVNF t"MMTPMJEJOHSFEJFOUTXFSFBEEFEUPUIFNJYJOH
XBTSFRVJSFEUPDPWFSUIFJNQFMMFS.JYJOHXBTJOJUJ tank
BUFEBOEPCTFSWFEUPQSPWJEFBWPSUFYUIBUXPVME t"GUFSBMMJOHSFEJFOUTXFSFBEEFE UIFTPMVUJPO
EJTTPMWFUIFJOHSFEJFOUT5IFIZESPQIJMJDQPMZNFS XBTNJYFEGPSNJOVUFT
XBTLOPXOUPCFUIFTMPXFTUEJTTPMWJOHPGBMMJOHSF t1VSJGJFEXBUFSXBTBEEFEUPUIFNJYJOHUBOLUP
EJFOUT5IFSFRVJSFEBNPVOUPGIZESPQIJMJDQPMZNFS PGUIFGJOBMWPMVNF
was added to the purified water in the tank. The t4PMVUJPOQ)XBTBEKVTUFEXJUIIZESPDIMPSJD
QPMZNFSXBTRVJDLMZESBXOJOUPUIFMJRVJEBOEFBT BDJETPMVUJPOPSTPEJVNIZESPYJEFTPMVUJPO JG
JMZEJTTPMWFE necessary
t1VSJGJFEXBUFSXBTBEEFETVGGJDJFOUUPNBLFUIF
Run Number Two GJOBMTPMVUJPOWPMVNF
"TFDPOEUSJBMSVOXBTDPOEVDUFEVTJOHBMMJOHSFEJFOUT t5IFTPMVUJPOXBTNJYFEGPSNJOVUFT
FYDFQUUIFBDUJWFQIBSNBDFVUJDBMJOHSFEJFOU/PU t4PMVUJPOXBTGJMUFSFEBOEGJMMFEJOUPUIFDPN
JODMVEJOHUIFBDUJWFJOHSFEJFOUDPOTJEFSBCMZSFEVDFE NFSDJBMQBDLBHJOH
UIFDPTUPGUIFUSJBMSVO1VSJGJFEXBUFSXBTBEEFE
UPUIFNJYJOHUBOLBUPGUIFGJOBMWPMVNF.JY Training
JOHXBTJOJUJBUFE"MMJOBDUJWFJOHSFEJFOUTXFSFUIFO "MMQFSTPOOFMXFSFUSBJOFEPOUIFOFXNBOVGBDUVS
BEEFEUPUIFQVSJGJFEXBUFS"MMJOHSFEJFOUTRVJDLMZ JOHQSPDFEVSF5IFJNQPSUBODFPGBEEJOHPOMZ
EJTTPMWFE5IJTTFDPOESVODPOGJSNFEUIBUBMMGPSNVMB PGUIFGJOBMTPMVUJPOWPMVNFUPUIFNJYJOHUBOLXBT
UJPOJOHSFEJFOUTXPVMECFEJTTPMWFEJOUIFTPMV stressed in the training. This procedure was used
UJPOWPMVNFEVSJOHUIFNJYJOHQSPDFTT5IFUSJBMSVO POMZXIFOVTJOHUIFUXPJNQFMMFSNJYJOHUBOL
HSFBUMZJODSFBTFEUIFDPOGJEFODFPGBMMJOWPMWFEUIBU
UIFQSPDFTTWBMJEBUJPOXPVMECFTVDDFTTGVM#FDBVTF Documentation
UIFBDUJWFJOHSFEJFOUXBTSFBEJMZTPMVCMFJOXBUFS JU "MMXPSLBTTPDJBUFEXJUIUIFPSJHJOBMWBMJEBUJPOXBT
XBTOPUFYQFDUFEUPJNQBDUUIFTPMVUJPOQSPDFTT DPNQMFUFE"MMJOWFTUJHBUJPOT BOBMZTFT BOEDPODMV
"OFXCBUDISFDPSEXBTEFWFMPQFEVTJOHUIF TJPOTXFSFEPDVNFOUFEUPDMPTFUIFPSJHJOBMGBJMFE
OFXMZEFWFMPQFENJYJOHQSPDFTT/FXWBMJEBUJPO WBMJEBUJPO%FWFMPQNFOUPGUIFOFXNBOVGBDUVSJOH

Paul L. Pluta
QSPDFTTJODMVEJOHUSJBMSVOTXBTEPDVNFOUFEBOE GSPNPDDVSSJOHXIFONPSFDPNQMFYGPSNVMBUJPOT
GJMFEJOUIFWBMJEBUJPOMJCSBSZBTTVQQPSUJOHJOGPS DPOUBJOJOHDFSUBJOJOHSFEJFOUTXFSFNBOVGBDUVSFE
NBUJPOGPSUIFOFXWBMJEBUJPO12SVOT
Other Similar Products Manufactured at
Process Validation the Site
5IFJODSFBTFECBUDITJ[FNBOVGBDUVSJOHQSPDFTT .BOVGBDUVSJOHCBUDISFDPSETGPSPUIFSQSPEVDUT
GPSQSPEVDUJOUIFCBDLVQUXPJNQFMMFSNBOVGBD XJUIMBSHFCBUDITJ[FTXFSFSFWJFXFE.BTUFSCBUDI
UVSJOHUBOLXBTWBMJEBUFE5ISFFMPUTXFSFUFTUFE records for certain products were noted to exclude
"MMUFTUEBUBQBTTFEBDDFQUBODFDSJUFSJB5IFOFX UIFUXPJNQFMMFSUBOLGSPNPQUJPOBMVTFJONBOV
NBOVGBDUVSJOHQSPDFTTXBTTVDDFTTGVMBOEUIF GBDUVSJOH"MMQSPEVDUTTQFDJGJFEJOJUJBMBEEJUJPO
QSPDFTTXBTWBMJEBUFE PGPGQVSJGJFEXBUFSUPUIFNJYJOHUBOL5IJT
WPMVNFXPVMEIBWFSFTVMUFEJOJOBEFRVBUFNJYJOHJG
Post-Validation Monitoring UIFUXPJNQFMMFSUBOLXBTVTFEGPSNBOVGBDUVSJOH
"EEJUJPOBMNPOJUPSJOHPGNBOVGBDUVSJOHUIF PGQSPEVDUTDPOUBJOJOHDFSUBJOGPSNVMBUJPOJOHSFEJ
JODSFBTFECBUDITJ[FQSPEVDUXBTDPOEVDUFEUPDPO FOUT*GUIFUXPJNQFMMFSUBOLXPVMECFOFFEFEGPS
GJSNTVDDFTTGVMQSPDFTTWBMJEBUJPO7JTVBMPCTFS NBOVGBDUVSJOHGMFYJCJMJUZPGDFSUBJOQSPEVDUT UIF
WBUJPOPGUIFNJYJOHBOEEJTTPMVUJPOQSPDFTTXBT NBTUFSCBUDISFDPSETPGUIFPUIFSQSPEVDUTXPVMECF
SFRVJSFEGPSTVCTFRVFOUMPUTNBOVGBDUVSFEEVSJOH DIBOHFEUPBEEPGQVSJGJFEXBUFSUPUIFNJYJOH
UIFOFYUUISFFNPOUIT0OFMPUXBTNBOVGBDUVSFE UBOLJOUIFGJSTUTUFQPGUIFNBOVGBDUVSJOHQSPDFTT
FBDINPOUI5IFNBOVGBDUVSJOHQSPDFTTVTJOHUIF
OFXQSPDFTTXBTTVDDFTTGVMJOBMMCBUDIFT7JTVBM POST CAPA
PCTFSWBUJPOPGUIFNBOVGBDUVSJOHQSPDFTTXBTEPD 4FWFSBMPUIFSQSPEVDUTXFSFMBUFSSFRVJSFEUPCF
VNFOUFEBOEGJMFEJOUIFWBMJEBUJPOMJCSBSZ NBOVGBDUVSFEJOUIFUXPJNQFMMFSUBOLJOUIFGPM
MPXJOHNPOUIT.BOVGBDUVSJOHFYQFSJFODFXJUI
Equipment Qualification and Equivalence the product of this case study including trial runs
5IFJOTUBMMBUJPORVBMJGJDBUJPO *2
GPSUIFNBOV XFSFVTFEBTTVQQPSUJOHKVTUJGJDBUJPOUPNPEJGZUIF
GBDUVSJOHUBOLXJUIUIFUXPJNQFMMFSTXBTSFWJFXFE SFTQFDUJWFNBTUFSCBUDISFDPSETPGUIFBEEJUJPOBM
UPCFTVSFUIBUJUTEPDVNFOUBUJPODPSSFDUMZTUBUFE QSPEVDUT5IFGPSNVMBUJPOTGPSBMMQSPEVDUTXFSF
JUTJNQFMMFSBSSBOHFNFOU5IF*2EPDVNFOUBUJPO FTTFOUJBMMZJEFOUJDBMFYDFQUJOHUIFBDUJWFJOHSFEJ
GPSBMMPUIFSUBOLTXBTBMTPSFWJFXFE8IJMFUIF FOU*OBMMDBTFT UIFBDUJWFJOHSFEJFOUXBTSFBEJMZ
SFTQFDUJWF*2EPDVNFOUTXFSFBDDVSBUF FRVJQNFOU TPMVCMFJOQVSJGJFEXBUFS.BOVGBDUVSJOHQSPDFTT
FRVJWBMFODFEPDVNFOUTJODPSSFDUMZJOEJDBUFEUIBU FTXFSFNPEJGJFE BOE12SVOTXFSFDPOEVDUFEUP
BMMUBOLTXFSFJOUFSDIBOHFBCMFCFDBVTFUIFZBMMXFSF WFSJGZUIFTVDDFTTPGUIFOFXNBOVGBDUVSJOHQSP
UIFTBNFTJ[FBOEWPMVNF&RVJQNFOUFRVJWBMFODF DFTT0OF12SVOXBTDPOEVDUFEGPSWBMJEBUJPO
EPDVNFOUTXFSFDIBOHFEUPJOEJDBUFUIBUUIFUXP
JNQFMMFSUBOLXBTOPUMJTUFEBTFRVJWBMFOUUPUIF CONCLUSIONS
PUIFSUISFFJNQFMMFSUBOLTGPSBMMQSPEVDUT8IJMF "DBTFTUVEZJOWPMWJOHUIFQSPDFTTWBMJEBUJPOPGBO
UIFNBOVGBDUVSJOHWPMVNFTPGBMMUBOLTXFSFJEFOUJ JODSFBTFECBUDITJ[FPGBOBRVFPVTMJRVJETPMVUJPO
DBM UIFNJYJOHDBQBCJMJUJFTPGUIFUBOLTXFSFTJHOJGJ QSPEVDUXBTEFTDSJCFE5IFCBUDITJ[FJODSFBTFXBT
DBOUMZEJGGFSFOUCFDBVTFUIFJNQFMMFSTPOUIFUBOLT DPOTJEFSFEUPCFBSFMBUJWFMZTJNQMFDIBOHFTJODF
were different. PUIFSTJNJMBSQSPEVDUTBUUIFTJUFXFSFBMSFBEZNBOV
5IFUXPJNQFMMFSBOEUISFFJNQFMMFSUBOLTJO GBDUVSFEBUUIFTBNFJODSFBTFECBUDITJ[F5IFOFFE
UIJTFYBNQMFXFSFPSJHJOBMMZEFTJHOBUFEBTFRVJWB UPBDUVBMMZEFNPOTUSBUFTVDDFTTGVMQSPDFTTJOHJO
MFOUNBOZZFBSTBHPXIFOUIFTJUFNBOVGBDUVSFE WBMJEBUJPOXBTFWFORVFTUJPOFECFDBVTFUIFQSPEVDU
SFMBUJWFMZTJNQMFGPSNVMBUJPOT5IFTFGPSNVMBUJPOT BOEQSPDFTTXFSFJEFOUJDBMUPPUIFSNBOVGBDUVSJOH
DPOUBJOFEWFSZTPMVCMFJOHSFEJFOUTUIBUEJTTPMWFE BOEUIFNJYJOHUBOLXBTUIPVHIUUPCFJEFOUJDBMUP
RVJDLMZJOBMMUBOLT"MMUBOLTQSPWJEFEFRVJWBMFOU BMMPUIFSUBOLT5IFSFXBTUIPVHIUUPCFOPQPTTJCJM
QFSGPSNBODFXIFOUIFTFTJNQMFGPSNVMBUJPOT OP JUZPGBOZQSPCMFNTPDDVSSJOH.BOVGBDUVSJOHXBT
TMPXEJTTPMWJOHQPMZNFST
XFSFNBOVGBDUVSFE BOE TUBSUFE BOEUIFVOFYQFDUFEIBQQFOFEJOHSFEJFOUT
UIFZXFSFUIVTEFFNFEUPCFFRVJWBMFOUBUUIBUUJNF EJEOPUEJTTPMWFJOUIFQSPEVDUTPMVUJPO BOEUIF
&RVJQNFOUFRVJWBMFODFEPDVNFOUBUJPOXBT NBOVGBDUVSJOHCBUDIXBTBDPNQMFUFGBJMVSF
NPEJGJFEUPJOEJDBUFUIBUBMMUBOLTDPVMECFVTFE 5IFDBVTFPGUIFGBJMVSFXBTFBTJMZEFUFSNJOFE
JOUFSDIBOHFBCMZGPSDFSUBJOGPSNVMBUJPOT CVU 5IFJNQFMMFSTPOUIFNJYFSJOUIFTQFDJGJDNJYJOH
XFSFOPUFRVJWBMFOUGPSPUIFSGPSNVMBUJPOT5IJT UBOLVTFEGPSUIFJODSFBTFECBUDITJ[FXFSFEJGGFSFOU
BQQSPBDIHBWFNBOVGBDUVSJOHNBYJNVNGMFYJCJMJUZ UIBOBMMPUIFSTJUFNJYJOHUBOLT.JYJOHJOUIJTUBOL
BOEQSFWFOUFEGVUVSFQPUFOUJBMNJYJOHQSPCMFNT XBTJOBEFRVBUF"OFXNBOVGBDUVSJOHQSPDFTTXBT

Paul L. Pluta
VMUJNBUFMZEFWFMPQFEUIBUFOBCMFEBEFRVBUFNJY 5IJTDBTFTUVEZEFNPOTUSBUFTUIFOFFEGPSDPNQMJ
JOHBOETVDDFTTGVMNBOVGBDUVSJOH"MMBTTPDJBUFE BODFQSPGFTTJPOBMTUPCFWJHJMBOUFWFOXIFOBQQBS
EPDVNFOUBUJPOBGGJSNJOHUIBUBMMNJYJOHUBOLTXFSF FOUMZNVOEBOFDIBOHFTBSFJOJUJBUFEi*EFOUJDBMw
FRVJWBMFOUXBTDPSSFDUFE NBZSFBMMZOPUCFTP TFF3FGFSFODF
"OPUIFSMFT
5IJTJODJEFOUSFNJOEFEBMMHSPVQTBUUIFTJUFUIBU TPO-PPLJOTJEFUIFFRVJQNFOU
UIFZNVTUCFTFOTJUJWFUPQFSDFJWFENJOPSEJGGFS
FODFTJOFRVJQNFOU5IFUXPJNQFMMFSBOEUISFF REFERENCE
JNQFMMFSUBOLTJOUIJTFYBNQMFXFSFPSJHJOBMMZEFT '%" i7BMJEBUJPOBOE&RVJQNFOU2VBMJGJDBUJPOo8IFO
JHOBUFEBTFRVJWBMFOUNBOZZFBSTBHPXIFOUIFTJUF A*EFOUJDBM3FBMMZ*TOU wHuman Drug CGMP Notes 7PM
NBOVGBDUVSFESFMBUJWFMZTJNQMFGPSNVMBUJPOT5IFTF VNF /VNCFS 4FQUFNCFSJVT
GPSNVMBUJPOTDPOUBJOFEWFSZTPMVCMFJOHSFEJFOUT
UIBUEJTTPMWFERVJDLMZJOBMMUBOLT"MMUBOLTQSP ARTICLE ACRONYM LISTING
WJEFEFRVJWBMFOUQFSGPSNBODFXIFOUIFTFTJNQMF CAPA Corrective Action and Preventive Action
GPSNVMBUJPOTXFSFNBOVGBDUVSFE)PXFWFS XIFO CGMP Current Good Manufacturing Practice
NPSFDPNQMFYGPSNVMBUJPOTSFRVJSJOHFOIBODFE GMP Good Manufacturing Practice
NJYJOHXFSFNBOVGBDUVSFE EJGGFSFODFTCFUXFFO IQ Installation Qualification
NJYJOHUBOLT JF UXPJNQFMMFSTWTUISFFJNQFMMFST
PQ Performance Qualification
CFDBNFBQQBSFOU QA Quality Assurance
0SJHJOBMMZQVCMJTIFEJOUIF4VNNFSJTTVFPGThe Journal of Validation Technology

Paul L. Pluta
Like-For-Like Changes: Is Validation

Testing Needed?
Validation Case Study #7
Paul L. Pluta
Validation Case Studies discusses validation situ- FRVJQNFOUSFTVMUFEJOTFSJPVTDPOTFRVFODFTGPS

ations useful to practitioners in validation and UIFNBOVGBDUVSJOHTJUF
compliance. Each case presented deals with a spe- t5IFTJUFWBMJEBUJPOBQQSPWBMDPNNJUUFF 7"$

cific validation problem, elements of which are TIPVMEFWBMVBUFBMMDIBOHFT JODMVEJOHMJLFGPS
described to demonstrate strategy to solve valida- MJLFDIBOHFT UPTUBOEBSEJ[FFWBMVBUJPOBOE
tion problems. We intend this column to be a use- EFDJTJPOQSPDFTTFT BOEUPDBSFGVMMZDPOTJEFS
ful resource for daily work applications. associated risks.
Reader comments, questions, and suggestions are t&NFSHFODZDIBOHFTNBZCFBQQSPWFEBTOFFEFE
needed to help us fulfill our objective for this col- CZNBJOUFOBODFPSPUIFSNBOBHFNFOU)PXFW
umn. Please send your comments and suggestions FS UIFTFDIBOHFTTIPVMEVMUJNBUFMZCFSFWJFXFE
to managing editor Susan Haigney at shaigney@ CZUIF7"$GPSGJOBMEJTQPTJUJPO
advanstar.com. t%PDVNFOUBUJPOBUUFTUJOHUPDPSSFDUJOTUBMMB
UJPOPGIJHISJTL MJLFGPSMJLFFRVJQNFOUDBOCF
KEY POINTS DISCUSSED BDDPNQMJTIFECZTJNQMFBQQSPWFENFNP5IJT
The following key points are discussed: BQQSPBDIQSPWJEFTDFSUBJOUZPGTVDDFTTGVMJOTUBM
t$IBOHFDPOUSPMJTBDSJUJDBMBDUJWJUZJOSFHVMBUFE MBUJPOXIJMFSFEVDJOHCVSEFOTPNFEPDVNFOUB
QIBSNBDFVUJDBMNBOVGBDUVSJOH UJPOSFRVJSFNFOUT
t$IBOHFTUPNBOVGBDUVSJOHQSPDFTTFT GPSNVMB t&OHJOFFSJOHBOENBJOUFOBODFBDUJWJUJFTTIPVME
UJPOT FRVJQNFOU DMFBOJOHQSPDFEVSFT BOBMZUJ IBWFDMFBSBOEEFGJOFEQSPDFEVSFTGPSMJLFGPS
DBMNFUIPET BOEPUIFSTZTUFNTBSFFWBMVBUFE MJLFFRVJQNFOUSFQMBDFNFOU1FSTPOOFMTIPVME
CBTFEPOSJTLPSJNQBDU"QQSPQSJBUFWBMJEBUJPO CFDBVUJPOFEBHBJOTUiDVUUJOHDPSOFSTwEVFUP
PSRVBMJGJDBUJPOXPSLJTUIFOQSFTDSJCFE CVTJOFTTQSFTTVSFT
ti-JLFGPSMJLFwDIBOHFTBSFVTVBMMZDPOTJEFSFE t7BMJEBUJPOBOEDPNQMJBODFQFSTPOOFMTIPVME
UPCFNJOJNBMDIBOHFTOPUSFRVJSJOHDPOGJSNB CFXBSZPGTVQQPTFEJODPOTFRVFOUJBMMJLFGPS
UPSZWBMJEBUJPOUFTUJOH MJLFDIBOHFTBOEFOTVSFUIBUTVDIFRVJQNFOU
tThis discussion addresses situations in which JOIJHISJTLBQQMJDBUJPOTJTQSPQFSMZJOTUBMMFE
MJLFGPSMJLFDIBOHFTEJEOPUQFSGPSNBTFYQFDU JODMVEJOHBQQSPQSJBUFEPDVNFOUBUJPO
ed due to incorrect installation of the replace
NFOUFRVJQNFOU4FWFSBMBDUVBMPDDVSSFODFT INTRODUCTION
BSFEFTDSJCFE'BVMUZJOTUBMMBUJPOPGMJLFGPSMJLF $IBOHFDPOUSPMJTBDSJUJDBMDPNQPOFOUJOSFHVMBUFE
[
ABOUT THE AUTHORS
For more Author Paul L. Pluta, Ph.D., is a pharmaceutical scientist with extensive technical and management
information, experience in the pharmaceutical industry. He is also adjunct assistant professor at the University of
go to Illinois at Chicago (UIC) College of Pharmacy in Chicago, Illinois USA. He may be contacted by e-mail
gxpandjvt.com/bios at paul.pluta@comcast.net.

Paul L. Pluta
QIBSNBDFVUJDBMNBOVGBDUVSJOHBOERVBMJUZTZTUFNT 1. Change desired."OFXQSPEVDUPSQSPDFTT QSP

$IBOHFTUPNBOVGBDUVSJOHQSPDFTTFT GPSNVMBUJPOT DFTTDIBOHF PSFRVJQNFOUDIBOHFJTOFFEFEPS
FRVJQNFOU DMFBOJOHQSPDFEVSFT BOBMZUJDBMNFUI SFRVJSFE5IJTNBZCFBOFDFTTBSZDIBOHF BEFTJS
PET BOEPUIFSTZTUFNTBSFFWBMVBUFECBTFEPOSJTLPS BCMFJNQSPWFNFOU FNFSHFODZXPSL PSSPVUJOF
JNQBDU$PNQMJDBUJOHUIFTFKVEHNFOUTBSFFNFSHFODZ NBJOUFOBODF
TJUVBUJPOTXIFODIBOHFTNVTUCFNBEFUPNBJOUBJO Support pre-work."QQSPQSJBUFQSFXPSLJTDPN
PQFSBUJPOTXJUIPVUQSPBDUJWFUIPSPVHIFWBMVBUJPOPG QMFUFEJOBEWBODFBOETVQQPSUJOHUIFDIBOHF
SJTLPSJNQBDU 5IJTXPSLNBZSBOHFGSPNIJHIMZDPNQMFYBOE
3JTLFWBMVBUJPOPGDIBOHFJTBDSJUJDBMEFUFSNJOBUJPO GBSSFBDIJOH FH OFXQSPDFTTEFWFMPQNFOU OFX
"UPOFFYUSFNF DIBOHFTNBZCFTJHOJGJDBOUSFQSF FRVJQNFOUQVSDIBTF GBDUPSZBDDFQUBODFUFTUJOH

senting a high risk to continuing critical processing or UPWFSZNJOJNBM FH TFMFDUJOHBOEPSEFSJOHB
PUIFSPQFSBUJPOT4VDIDIBOHFTSFRVJSFGVMMWBMJEBUJPO NJOPSSFQMBDFNFOUQBSU

PSRVBMJGJDBUJPO"UUIFPUIFSFYUSFNF DIBOHFTNBZCF Validation or change request. " GPSNBM
DPNQMFUFMZJOTJHOJGJDBOUXJUINJOJNBMSJTLUPQBUJFOU SFRVFTUUPJOJUJBUFUIFWBMJEBUJPOQSPDFTTJTTVC
QSPEVDU QSPDFTT PSFRVJQNFOUi-JLFGPSMJLFwDIBOHFT NJUUFE UP UIF WBMJEBUJPO BQQSPWBM DPNNJUUFF
BSFPGUFOKVEHFEUPCFDIBOHFTJOUIJTMBUUFSDBUFHPSZ 7"$
4VQQPSUXPSLSFQPSUTNBZCFJODMVEFEJO
-JLFGPSMJLFDIBOHFTBSFNJOJNBMDIBOHFTNBEFUP UIFSFRVFTUJOTVQQPSUPGUIFDIBOHF5IFDIBOHF
QSPDFTTFT FRVJQNFOU VUJMJUJFT GBDJMJUJFT DPNQVUFST PS SFRVFTUNBZJODMVEFBQSPQPTFEMFWFMPGXPSLUP
PUIFSTZTUFNT-JLFGPSMJLFDIBOHFTBSFVTVBMMZKVEHFE DPOGJSNUIFBDDFQUBCJMJUZPGUIFDIBOHF5IFMFWFM
UPCFFRVJWBMFOUSFQMBDFNFOUT0GUFOUIFTFDIBOHFT PGXPSLJTCBTFEPOSJTLUPUIFQBUJFOUBOEUPUIF
BSFJEFOUJDBMSFQMBDFNFOUTPGUIFPSJHJOBMJUFNPGJOUFS PSHBOJ[BUJPO5IF7"$BQQSPWFTUIFWBMJEBUJPO
FTU4PNFNBZOPUDPOTJEFSMJLFGPSMJLFDIBOHFTUPCF SFRVFTU
BOBDUVBMDIBOHF"SFQMBDFNFOUNPUPS SFQMBDFNFOU Validation plan.%FQFOEJOHPOUIFTDPQFBOE
TDSFFO SFQMBDFNFOUGJMUFSCBH SFQMBDFNFOUQJQJOH BOE NBHOJUVEFPGUIFWBMJEBUJPOSFRVFTU BWBMJEB
TPPOBSFPGUFOSPVUJOFNBJOUFOBODFPOWBMJEBUFEPS UJPOQMBONBZCFTVCNJUUFE5IJTQMBOEFTDSJCFT
RVBMJGJFEFRVJQNFOU BOEOPUSFBMMZBOBDUVBMDIBOHF UIFWBMJEBUJPOXPSLSFRVJSFEJOHSFBUFSEFUBJM
-JLFGPSMJLFDIBOHFTBSFUIPVHIUUPOPUSFRVJSFBOZ BQQSPQSJBUFUFTUJOHBOETBNQMJOHQSPQPTFEUP
TVCTFRVFOUUFTUJOH5ZQJDBMMPHJDJTUIBUCFDBVTFUIF EFNPOTUSBUFBTVDDFTTGVMDIBOHF BOEBDDFQUBODF
PSJHJOBMFRVJQNFOUPSTZTUFNIBTBMSFBEZCFFOWBMJ DSJUFSJBGPSUIFSFDPNNFOEFEUFTUJOH5IF7"$
EBUFE BOECFDBVTFOPUIJOHIBTSFBMMZDIBOHFE XIZ BQQSPWFTUIFWBMJEBUJPOQMBO
UIFOFFEUPEPOFXWBMJEBUJPOXPSL 8IZXBTUFUJNF Validation protocol."QSPUPDPMJTXSJUUFOTQFDJ
PSFGGPSUXIFOOPUIJOHIBTDIBOHFE *TOUUIBUUIF GZJOHEFUBJMFETBNQMJOHBOEUFTUJOHUPDPOGJSNUIF
QSPCMFNXJUIWBMJEBUJPOoEPJOHUPPNVDIVOOFDFTTBSZ BDDFQUBCJMJUZPGUIFDIBOHF5IF7"$BQQSPWFTUIF
XPSL WBMJEBUJPOQSPUPDPM
5IJTEJTDVTTJPOBEESFTTFTTJUVBUJPOTJOXIJDIMJLF Validation work.7BMJEBUJPOXPSLJTQFSGPSNFE
GPSMJLFDIBOHFTEJEOPUQFSGPSNBTFYQFDUFE4FWFSBM BDDPSEJOHUPUIFQSPUPDPM4BNQMJOHBOEUFTUJOH
WBMJEBUJPONBOBHFSTQSFTFOUFEFYBNQMFTPGTVDITJUV BSFDPNQMFUFE%BUBBOEPUIFSSFTVMUTBSFHFOFS
BUJPOTBUJOUFSOBUJPOBMWBMJEBUJPODPOGFSFODFT5IF ated and recorded.
PDDVSSFODFTEFTDSJCFEBSFBDUVBMJODJEFOUT&YBNQMFT Validation report."SFQPSUDPOUBJOJOHBMMUFTU
EFTDSJCFEEFNPOTUSBUFUIFCSPBETDPQFBOETJHOJGJDBOU results with discussion and conclusions is pre
NBHOJUVEFPGQPUFOUJBMQSPCMFNTXJUIMJLFGPSMJLF QBSFEBOETVCNJUUFEUPUIF7"$GPSBQQSPWBM5IF
changes. SFQPSUJTBQQSPWFE BOEUIFQSPDFTTPSFRVJQNFOU
The following discussion includes: DIBOHFJTJNQMFNFOUFE"OBQQSPQSJBUFMFWFMPG
tBackground. 5IFUZQJDBMWBMJEBUJPOPSRVBMJGJDB QPTUWBMJEBUJPONPOJUPSJOHJTQSPQPTFEJOUIF
UJPOQSPDFTTJTEFTDSJCFE report.
tCompliance events.%FTDSJQUJPOTPGTFWFSBM Validation closure. If no other work is need
BDUVBMMJLFGPSMJLFWBMJEBUJPOFWFOUT FE UIFWBMJEBUJPOQSPKFDUJOJUJBUFECZUIFDIBOHF
tWhat are the issues?8IBUIBQQFOFEUPDBVTF SFRVFTUJTDMPTFE
UIFTFQSPCMFNT 9. Emergency changes.*ODBTFTPGFNFSHFODZ
tCorrective action and preventive action DIBOHFT NBOBHFNFOUNBZBVUIPSJ[FDIBOHFT
(CAPA).3FDPNNFOEFEBDUJPOTBOEJNQSPWF UPCFDPNQMFUFEBTOFFEFEBQBSUGSPNUIFBCPWF
NFOUTUPQSFWFOUGVUVSFQSPCMFNPDDVSSFODFT EFTDSJCFEQSPDFTT&NFSHFODZDIBOHFTTIPVMECF
FWBMVBUFECZUIF7"$BTTPPOBTQPTTJCMF4VQQPSU
BACKGROUND JOHEPDVNFOUBUJPOJTUIFODPNQMFUFEUPDPOGJSN
5IFQSPDFTTPGWBMJEBUJPOPSRVBMJGJDBUJPOUZQJDBMMZ UIFBDDFQUBCJMJUZPGUIFDIBOHFBOETVCNJUUFEUP
DPNQSJTFTUIFGPMMPXJOHTFRVFODFPGBDUJWJUJFT UIF7"$GPSBQQSPWBM

Paul L. Pluta
Documentation."MMSFMFWBOUEPDVNFOUBUJPO UJPOPGBMJLFGPSMJLFSFQMBDFNFOUCMPXFSNPUPSCZ
EFTDSJCFE BDUJWJUJFT
JODMVEJOHTVQQPSUJOH UIFTJUFNBJOUFOBODFTIPQTFWFSBMNPOUITQSJPS
QSFXPSLTIPVMECFTUPSFEJOUIFWBMJEBUJPOMJCSBSZ )&1"GJMUSBUJPOPGSPPNBJSIBEUIVTCFFODPN
PSSFBEJMZBDDFTTJCMFUISPVHIBLOPXMFEHFNBO QSPNJTFEGPSTFWFSBMNPOUIT EVSJOHXIJDIUJNF
BHFNFOUTZTUFN NVMUJQMFQSPEVDUMPUTXFSFNBOVGBDUVSFE5IFTFMPUT
11. Continued verification.1PTUWBMJEBUJPONPO had passed all product testing and were released to
JUPSJOHDPOGJSNJOHBDDFQUBCJMJUZPGUIFDIBOHF DPNNFSDJBMEJTUSJCVUJPO
continues throughout the lifecycle. 5IF64'PPEBOE%SVH"ENJOJTUSBUJPOXBTOPUJ
The issue addressed in this case study occurred GJFE"MMMPUTNBOVGBDUVSFEEVSJOHUIFUJNFQFSJPE
JOTUFQOVNCFSUISFF UIFWBMJEBUJPOPSDIBOHF XIFOUIFCMPXFSNPUPSXBTPQFSBUJOHJODPSSFDUMZ
SFRVFTU XIJDIJOJUJBUFTUIFGPSNBMEPDVNFOUFE XFSFSFDBMMFE"TJEFGSPNUIJTJODJEFOUCFJOHIJHIMZ
WBMJEBUJPOQSPDFTT&NFSHFODZDIBOHFTNBZBMTP FNCBSSBTTJOHUPUIFPSHBOJ[BUJPO UIFSFDBMMDPTUT
IBWFBQQMJDBUJPOJOUIJTEJTDVTTJPO4QFDJGJDBMMZ JOWPMWFEXFSFTJHOJGJDBOU
DIBOHFTSFRVFTUFEXFSFKVEHFEUPCFMJLFGPSMJLF
DIBOHFTUIBUTIPVMEIBWFIBEOPJNQBDUPOUIF Milling Solids for Tablet Products
QSPDFTTPSFRVJQNFOU#FDBVTFOPJNQBDUXBT "OPUIFSNBOBHFSEFTDSJCFEBOJODJEFOUJOXIJDI
FYQFDUFE OPUFTUJOHPSPUIFSXPSLXBTSFDPN NBKPSQSPDFTTJOHQSPCMFNTXJUITFWFSBMUBCMFU
NFOEFECZUIF7"$ PSCZNBOBHFNFOU
BTQBSU QSPEVDUTPDDVSSFE5IJTFWFOUPDDVSSFEJOBMBSHF
PGUIFDIBOHF-JLFGPSMJLFSFQMBDFNFOUXPSL NBOVGBDUVSJOHGBDJMJUZUIBUQSPEVDFENVMUJQMF
XBTBVUIPSJ[FEBOEDPNQMFUFE8PSLXBTEPDV TPMJEEPTBHFGPSNT FH UBCMFUT DBQTVMFT HSBOVMFT
NFOUFEPOFRVJQNFOUIJTUPSZMPHT NBJOUFOBODF QPXEFST
0DDBTJPOBMQSPEVDUQSPDFTTJOHQSPCMFNT
SFDPSET PSPUIFSTZTUFNTJOBDDPSEBODFXJUITJUF XFSFPCTFSWFEXJUITFWFSBMQSPEVDUT"TUIFTFJOEJ
procedures. WJEVBMJODJEFOUTCFDBNFNPSFGSFRVFOU JOWFTUJHB
UJPOTXFSFJOJUJBUFE5IFQSPCMFNXBTFWFOUVBMMZ
VALIDATION EVENTS BTTPDJBUFEXJUIQBSUJDMFTJ[FEJTUSJCVUJPODIBOHFT
"EJTDVTTJPOXJUITFWFSBMWBMJEBUJPONBOBHFST #MFOEFEHSBOVMBUJPOTIBETNBMMFSQBSUJDMFTJ[FEJT
SFHBSEJOHQSPCMFNTXJUIWBMJEBUJPOBDUJWJUJFTJOEJ USJCVUJPOTUIBOFYQFDUFEXIFODPNQBSFEUPIJTUPSJ
DBUFEUIBUMJLFGPSMJLFDIBOHFTXFSFBSFDVSSJOHQSPC DBMEBUB&WFOUVBMMZUIFTFQSPCMFNTXFSFBTTPDJBUFE
MFN*OHFOFSBM DIBOHFTBSFEFGJOFEBTMJLFGPSMJLFUP XJUIQSPDFTTJOHVTJOHBTQFDJGJDJNQBDUNJMM8IFO
FMJNJOBUFBEEJUJPOBMWBMJEBUJPOXPSL5IFEFGJOJUJPO PQFSBUJPOPGUIFNJMMXBTJOWFTUJHBUFE JUXBTGPVOE
PGMJLFGPSMJLFNBZCFCSPBEMZJOUFSQSFUFEJOUIFTF UIBUUIFNJMMXBTSVOOJOHJOUIFSFWFSTFEJSFDUJPO
DBTFTBOEJOTPNFDBTFT UPPCSPBEMZJOUFSQSFUFE DPNQBSFEUPPUIFSNJMMTBUUIFTJUF5IJTDBVTFE
0DDBTJPOBMMZ UIFTFKVEHNFOUTBOEFOTVJOHQSPC JODPSSFDUNJMMJOHPGTPMJETBOEJNQSPQFSQBSUJDMF
MFNTIBETJHOJGJDBOUDPOTFRVFODFT4FWFSBMBDUVBM TJ[FEJTUSJCVUJPOT
FYBNQMFTEFTDSJCFECZNBOBHFSTEFNPOTUSBUFETVDI &RVJQNFOUIJTUPSZSFDPSETXFSFSFWJFXFE5IF
FWFOUT1FSTPOOFMSFTQPOTJCMFGPSUIFTFMJLFGPSMJLF POMZBDUJWJUZQFSGPSNFEUIBUDPVMEIBWFDBVTFEUIJT
KVEHNFOUTEJEOPUSFBMJ[FUIFQPUFOUJBMQSPCMFNT FWFOUXBTUIFJOTUBMMBUJPOPGBMJLFGPSMJLFSFQMBDF
DBVTFECZUIFJSKVEHNFOUT NFOUNPUPS5IFNPUPSDIBOHFXBTDPSSFMBUFEUPUIF
GJSTUSFQPSUPGQSPDFTTQSPCMFNTBUUIFTJUF1SPCMFNT
Laminar Air Flow in XFSFBTTPDJBUFEXJUINJMMJOHPGMBSHFRVBOUJUJFTPG
Aseptic Processing Facility ESVHBOEJOBDUJWFJOHSFEJFOUTGPSSFTQFDUJWFMPUT
0OFNBOBHFSEFTDSJCFEBOJODJEFOUJOXIJDI
SPVUJOFSFWBMJEBUJPOPGBOBTFQUJDGJMMJOHMJOFXBT Cell Culture Agitation
CFJOHQFSGPSNFE"MMBDUJWJUJFTBOEUFTUTXFSFQFS "OPUIFSWBMJEBUJPONBOBHFSEFTDSJCFEBOFWFOUJO
GPSNFEBDDPSEJOHUPQSPDFEVSF JODMVEJOHTUFSJM XIJDIUIFDFMMHSPXUIJOBCJPUFDIOPMPHZQSPDFTT
J[FSSFWBMJEBUJPO NFEJBSVOT JOUFSWFOUJPOT BOE XBTTJHOJGJDBOUMZMPXFSUIBOUIFIJTUPSJDBMBWFSBHF
TPPO"MMSFTVMUTXFSFBDDFQUBCMFBOEBTFYQFDUFE XIFODVMUVSFTXFSFHSPXOJOBTQFDJGJDNBOV
VOUJMUIFGJMMJOHMJOFTNPLFUFTUXBTDPOEVDUFE facturing tank. Because this product was a high
5IFTNPLFUFTUDPOGJSNTUIFMBNJOBSBJSGMPXEJSFD WPMVNFQSPEVDUUIBUXBTNBEFJONVMUJQMFUBOLT
UJPOJOUIFBTFQUJDTVJUF"MMJOWPMWFEXFSFTIPDLFE UIFMPXFSHSPXUIXBTOPUJNNFEJBUFMZPCWJPVT"T
XIFOUIFBJSGMPXXBTTIPXOUPCFVQXBSEUPXBSE NPSFMPXHSPXUICBUDIFTXFSFNBOVGBDUVSFE JU
UIF)&1"GJMUFSJOTUFBEPGEPXOXBSEUISPVHIUIF CFDBNFBQQBSFOUUIBUMPXZJFMETXFSFCFJOHDPOTJT
filter toward open product. tently produced in a single tank.
3PPNBOEFRVJQNFOUIJTUPSZSFDPSETXFSF &RVJQNFOUIJTUPSZSFDPSETGPSUIFTVTQFDUUBOL
DIFDLFE5IFPOMZBDUJWJUZQFSGPSNFEJOUIFBSFB XFSFSFWJFXFE5IFPOMZXPSLQFSGPSNFEPOUIF
UIBUDPVMEIBWFDBVTFEUIJTFWFOUXBTUIFJOTUBMMB UBOLBTEPDVNFOUFEJOUIFFRVJQNFOUIJTUPSZMPH

Paul L. Pluta
XBTUIFJOTUBMMBUJPOPGBSFQMBDFNFOUNPUPSPO DPOWFZPSJTHPJOHCBDLXBSET UIFQSPCMFNXPVME

UIFNJYFS5IFNPUPSDIBOHFXBTDPSSFMBUFEUP CFFBTJMZPCTFSWFE5IFPOMZMJLFGPSMJLFDIBOHFT
UIFGJSTUSFQPSUPGMPXFSFEHSPXUIBOETVCTFRVFOU SFRVJSJOHWFSJGJDBUJPOBSFUIPTFJOXIJDIJOTUBMMB
CBUDIFTUIFSFBGUFS8IFOPQFSBUJPOPGUIFUBOL UJPOFSSPSTXPVMEOPUCFPCWJPVTEVSJOHPQFSBUJPO
XBTUIPSPVHIMZJOWFTUJHBUFE JUXBTGPVOEUIBU FH SFWFSTFEMBNJOBSBJSGMPX

UIFJNQFMMFSXBTSVOOJOHJOUIFSFWFSTFEJSFDUJPO *OTPNFPSHBOJ[BUJPOT NBJOUFOBODFNBOBHF
DPNQBSFEUPUIFPSJHJOBMJOTUBMMBUJPOBOEDPN NFOUIBTUIFGJOBMBOEDPNQMFUFBVUIPSJUZUPKVEHF
pared to other tanks at the site. This effect signifi XIFUIFSPSOPUWBMJEBUJPOPGBDIBOHFVOEFSUIFJS
DBOUMZBMUFSFEUIFNJYJOHQSPGJMFPGUIFDFMMDVM BVTQJDFTJTSFRVJSFE5IJTQSBDUJDFTIPVMECFMJN
UVSF SFTVMUJOHJOSFEVDFEDFMMHSPXUIBOEBDUJWF JUFEUPJOJUJBUJPOPGFNFSHFODZDIBOHFTPOMZ*GB
QIBSNBDFVUJDBMJOHSFEJFOUZJFME DIBOHFJTMJLFGPSMJLFJOUIFKVEHNFOUPGNBJOUF
OBODFNBOBHFNFOU UIFFNFSHFODZXPSLTIPVME
WHAT ARE THE ISSUES? CFJOJUJBUFEBTTPPOBTQPTTJCMF8PSLTIPVMEUIFO
*OFBDIPGUIFFWFOUTQSFTFOUFE BMJLFGPSMJLF CFEPDVNFOUFEJOFRVJQNFOUIJTUPSZMPHTPSPUIFS
DIBOHFXBTQFSGPSNFEXJUIPVUOFFEGPSBOZDPO TJUFEPDVNFOUBUJPOTZTUFNT$PNQMFUFEFNFSHFODZ
GJSNBUPSZUFTUJOH*EFOUJDBMSFQMBDFNFOUFRVJQNFOU DIBOHFTTIPVMEUIFOCFFWBMVBUFECZUIFTJUF7"$
XBTBWBJMBCMFGPSFBDIPGUIFDIBOHFT"MMQFSTPOOFM $PNQMFUFEDIBOHFTBSFUIFOFWBMVBUFECBTFEPO
JOWPMWFECFMJFWFEUIBUJOTUBMMBUJPOPGSFQMBDFNFOU XPSLEPOFBOESJTLUPQBUJFOUBOEPSHBOJ[BUJPO
FRVJQNFOUXBTJODPOTFRVFOUJBMBOEXPVMECFEPOF "QQSPQSJBUFDPOGJSNBUPSZXPSLXPVMEUIFOCF
DPSSFDUMZ5IFQSPCMFNTPCTFSWFEXFSFDPNQMFUFMZ QSFTDSJCFE.BJOUFOBODFNBOBHFNFOUTIPVMECF
VOFYQFDUFE*OFBDIDBTF UIFMJLFGPSMJLFDIBOHF QFSNJUUFEUPBVUIPSJ[FFNFSHFODZDIBOHFT"MM
XBTDPOTJEFSFEUPCFJOTJHOJGJDBOU"EEJUJPOBMUFTU DIBOHFTVMUJNBUFMZTIPVMECFSFWJFXFE FWBMVBUFE
JOHUPWFSJGZBDDFQUBCMFFRVJQNFOUJOTUBMMBUJPOXBT BOEKVEHFECZUIFTJUF7"$5IF7"$KVEHNFOUTBSF
OPUQSFTDSJCFE"MUIPVHIUIFMJLFGPSMJLFFRVJQ EPDVNFOUFEJOUIFVTVBMNBOOFS"DDFQUBCJMJUZPG
NFOUTFMFDUJPOXBTBDDVSBUFBOEBDDFQUBCMF JOTUBM DIBOHFTJTUIFVMUJNBUFSFTQPOTJCJMJUZPGUIF7"$
MBUJPOPGUIFFRVJQNFOUXBTOPU'BVMUZJOTUBMMBUJPO
SFTVMUFEJOTFSJPVTDPOTFRVFODFTGPSUIFNBOV Validation Documentation Requirements
GBDUVSJOHTJUFBOEUIFPSHBOJ[BUJPO*OUIFBTFQUJD 1SPUPDPMTTIPVMEOPUBMXBZTCFSFRVJSFEUPDPOGJSN
)&1"GJMUFSJODJEFOU BTJHOJGJDBOUSFDBMMPGQSPEVDU BDDFQUBCJMJUZPGBMJLFGPSMJLFJOTUBMMBUJPO"TJNQMF
SFMFBTFEUPUIFNBSLFUQMBDFXBTSFRVJSFE XSJUUFONFNPXJUIBQQSPQSJBUFSFWJFXBOEBQQSPW
BMBUUFTUJOHUPDPSSFDUJOTUBMMBUJPOPGSFQMBDFNFOU
CORRECTIVE AND FRVJQNFOUTIPVMECFTVGGJDJFOUUPDPNQMFUFUIF
PREVENTIVE ACTIONS XPSL5IFNFNPTIPVMEJODMVEFBTUBUFNFOUUIBU
5IFJODJEFOUTQSFTFOUFEIFSFJODBVTFEFBDIWBMJ DPSSFDUFRVJQNFOUPQFSBUJPOIBECFFODPOGJSNFE
EBUJPONBOBHFSUPNBLFDIBOHFTJOUIFDIBOHF CZBTFDPOEJOEJWJEVBM.BOBHFNFOUPGUIFSFTQPO
DPOUSPMSFWJFXQSPDFTTBUUIFJSSFTQFDUJWFNBOVGBD TJCMFHSPVQ FH NBJOUFOBODF FOHJOFFSJOH PUIFST

UVSJOHTJUFT4PNFPGUIFJSDPNNFOUTBOETVHHFTUFE BOEUIFTJUFRVBMJUZBTTVSBODF 2"
EFQBSUNFOU
BQQSPBDIFTBSFEFTDSJCFEBTGPMMPXT TIPVMEBQQSPWFUIFNFNP5IFBQQSPWFENFNP
XPVMEUIFOCFGJMFEXJUIUIFWBMJEBUJPOSFRVFTUUP
Validation Approval DMPTFUIFMJLFGPSMJLFBDUJWJUZ5IJTEPDVNFOUBUJPO
Committee Review of All Changes BQQSPBDIBDDPNQMJTIFTUIFOFFEUPWFSJGZIJHISJTL
7BMJEBUJPONBOBHFSTBHSFFEUIBUBMMDIBOHFT JOTUBMMBUJPOXJUIPVUUIFFODVNCSBODFPGWBMJEBUJPO
JODMVEJOHMJLFGPSMJLFDIBOHFT TIPVMEVMUJNBUFMZ QSPUPDPMT FYDFTTJWFBQQSPWBMT BOEPUIFSCVSEFOT
CFFWBMVBUFECZUIFTJUF7"$)BWJOHBMMDIBOHFT 4PNFNBOBHFSTIBWFTVDDFTTGVMMZJNQMFNFOUFE
SFWJFXFECZBTJOHMFSFTQPOTJCMFHSPVQ 7"$
TUBO UIJTBQQSPBDIGPSDFSUBJOIJHISJTLDIBOHFT/P
EBSEJ[FTFWBMVBUJPOBOEEFDJTJPOQSPDFTTFT$PO BEEJUJPOBMWFSJGJDBUJPOEPDVNFOUBUJPO PUIFSUIBO
TJEFSBUJPOGPSSJTLUPQBUJFOUBOEPSHBOJ[BUJPOBOE FRVJQNFOUIJTUPSZ FUD
XBTSFRVJSFEGPSJOTJHOJGJ
FOTVJOHKVEHNFOUTTIPVMECFVOJGPSNMZDPOEVDUFE DBOUMJLFGPSMJLFDIBOHFTPSGPSDIBOHFTJOXIJDI
CZUIF7"$"QQSPQSJBUFDPOGJSNBUJPOUIBUMJLF DPSSFDUJOTUBMMBUJPOXBTBQQBSFOUCZPCWJPVTGBVMUZ
GPSMJLFFRVJQNFOUJTQSPQFSMZJOTUBMMFEXPVMEUIFO FRVJQNFOUPQFSBUJPO
CFCBTFEPO7"$KVEHNFOU*ONBOZDBTFT BEEJ
UJPOBMDPOGJSNBUPSZXPSLXPVMEOPUCFSFRVJSFE Procedures and Training
'PSFYBNQMF WFSJGJDBUJPOPGBDPSSFDUMZJOTUBMMFE 5IFBCPWFBQQSPBDIFTUPEFDJTJPONBLJOHBOE
SFQMBDFNFOUDPOWFZPSCFMUNPUPSJTOPUOFDFT EPDVNFOUBUJPOTIPVMECFTQFDJGJFEJOBQQSPWFE
TBSZJGUIFNPUPSXBTJOTUBMMFEJODPSSFDUMZBOEUIF QSPDFEVSF4QFDJGZJOHUIFBDDFQUBCJMJUZPGTVDI

Paul L. Pluta
DPOGJSNBUPSZEPDVNFOUBUJPOQSPWJEFTGMFYJCJMJUZ MBSGBVMUZJOTUBMMBUJPOJODJEFOUTBGUFSJNQMFNFOUB
UPBDDPNQMJTIWFSJGJDBUJPOXJUIPVUUIFCVSEFOPG UJPOPGUIFOFXWFSJGJDBUJPOQSPDFEVSF
FYDFTTJWFQSPUPDPMSFRVJSFNFOUT"MMBTTPDJBUFE
JOEJWJEVBMTJOUFDIOJDBMGVODUJPOT WBMJEBUJPOQFS CONCLUSIONS
TPOOFM NBJOUFOBODFNBOBHFNFOU 7"$NFNCFST 5IJTDBTFTUVEZEFTDSJCFTTFWFSBMFYBNQMFTPG
BOE2"QFSTPOOFMTIPVMECFUSBJOFEPOUIFOFX JNQSPQFSJOTUBMMBUJPOPGMJLFGPSMJLFFRVJQNFOUUIBU
procedure. DBVTFETJHOJGJDBOUQSPCMFNTGPSUIFNBOVGBDUVSJOH
TJUF-JLFGPSMJLFDIBOHFTBSFHFOFSBMMZBTTVNFEUP
Equipment Installation CFPGOPDPOTFRVFODF)PXFWFS JNQSPQFSJOTUBMMB
and Maintenance Procedures UJPOPGTVDIFRVJQNFOUDBOIBWFTFSJPVTSBNJGJDB
7BMJEBUJPONBOBHFSTDPNNFOUFEUIBUFOHJOFFSJOH UJPOTJODMVEJOHQSPEVDUSFDBMMT7BMJEBUJPONBOBH
BOENBJOUFOBODFBSFBTNVTUIBWFDMFBSQSPDFEVSFT FSTBHSFFEUIBUBMMDIBOHFT JODMVEJOHMJLFGPSMJLF
GPSMJLFGPSMJLFFRVJQNFOUSFQMBDFNFOU1SPDFEVSFT DIBOHFT TIPVMECFFWBMVBUFECZUIFTJUF7"$UP
NVTUCFNBJOUBJOFEBOETUSFOHUIFOFEBTOFFEFE TUBOEBSEJ[FFWBMVBUJPOBOEEFDJTJPOQSPDFTTFT
XIFOJODJEFOUTPDDVS/FXQFSTPOOFMTIPVMECF %PDVNFOUBUJPOSFRVJSFNFOUTBUUFTUJOHUPDPSSFDU
NBEFBXBSFPGFWFOUTTVDIBTUIPTFQSFTFOUFEIFSF JOTUBMMBUJPOPGIJHISJTL MJLFGPSMJLFFRVJQNFOU
JOBOECFDBVUJPOFEBCPVUDVUUJOHDPSOFST#VTJOFTT DBOCFBDDPNQMJTIFECZBTJNQMFBQQSPWFENFNP
BOEPQFSBUJPOQSFTTVSFTPGUFOEFNBOESBQJEXPSL UIBUJTGJMFEXJUIUIFEPDVNFOUBUJPOSFRVFTU5IJT
GSPNUIFTFHSPVQT1FSTPOOFMNVTUVOEFSTUBOEUIBU BQQSPBDIQSPWJEFTDFSUBJOUZPGTVDDFTTGVMJOTUBM
QSPQFSFRVJQNFOUJOTUBMMBUJPONVTUOFWFSCFQPUFO MBUJPOXJUIPVUVOEVFEPDVNFOUBUJPOCVSEFO
UJBMMZDPNQSPNJTFE &OHJOFFSJOHBOENBJOUFOBODFQSPDFEVSFTGPSMJLF
GPSMJLFFRVJQNFOUSFQMBDFNFOUNVTUCFDMFBS7BMJ
Senior Management Support EBUJPOBOEDPNQMJBODFQFSTPOOFMTIPVMECFXBSZPG
7BMJEBUJPONBOBHFSTSFQPSUFEUIBUUIFJSSFTQFDUJWF TVQQPTFEJODPOTFRVFOUJBMMJLFGPSMJLFDIBOHFTBOE
TJUFTFOJPSNBOBHFNFOUXBTTVQQPSUJWFPGUIFBCPWF FOTVSFUIBUTVDIFRVJQNFOUJOIJHISJTLBQQMJDB
BQQSPBDIFT7BMJEBUJPONBOBHFSTSFQPSUFEOPTJNJ tions is properly installed. JVT

Understanding and Reducing

Analytical Error
Why Good Science Requires
Operational Excellence
Analysis and Control of Variation is dedicated to t"DUVBMFYBNQMFTPGWBSJBCMFQSPDFTTFTBSF

revealing weaknesses in existing approaches to under- presented.
standing, reducing, and controlling variation and to ti4UBCJMJ[FGJSTUwJTUIFGJSTUQSJODJQMF4UBCMFQSP
recommend alternatives that are not only based on DFTTFTBSFQSFEJDUBCMF
sound science but also that demonstrably work. Case t7BSJBUJPOJOMBCPSBUPSZPQFSBUJPOTNBZNBTL
studies will be used to illustrate both problems and causal relationships in other areas.
successful methodologies. The objective of the col- t$PNQMJBODFUPQSPDFEVSFTJTOPUBDDFQUBCMF
umn is to combine sound science with proven practical SBUJPOBMFGPSBWBSJBCMFQSPDFTT
advice. t4FOJPSNBOBHFNFOUTIPVMESFNPWFPCTUBDMFT
Reader comments, questions, and suggestions will UPDPORVFSJOHWBSJBUJPOCZNBLJOHJUBTUSBUFHJD
help us fulfil our objective for this column. Case studies JNQFSBUJWF
illustrating the successful reduction or control of varia- t*OFOWJSPONFOUTXIFSFIJHIEFHSFFTPGWBSJB
tion submitted by readers are most welcome. We need UJPOBSFQPTTJCMF FH JOCJPMPHJDT
UIFOFFE
your help to make Analysis and Control of Variation GPSWFSZMPXMFWFMTPGWBSJBUJPOJOPQFSBUJPOTJT
a useful resource. Please send your comments and greatest.
suggestions to column coordinator John McConnell at t3FEVDFEWBSJBUJPONFBOTGFXFSEFWJBUJPOT
john@wysowl.com.au or journal coordinating editor GFXFSSFTPVSDFTUJFEVQDPOEVDUJOHJOWFTUJHB
Susan Haigney at shaigney@advanstar.com. UJPOTBOESFQPSUT NPSFSFTPVSDFTEFEJDBUFE
UPEPJOHUIFDPSFXPSL BOEJODSFBTFETFDVSJUZ
KEY POINTS DISCUSSED GSPNSPCVTUQSPDFTTFTXJUILOPXODBQBCJMJUJFT
The following key points are discussed: t0QFSBUJOHJOBMPXWBSJBUJPOFOWJSPONFOU
t(PPETDJFODFJOEJTDPWFSZ EFWFMPQNFOU QSP results in easier detection of causal relation
EVDUJPO BOEJOMBCPSBUPSJFTSFRVJSFTTUBCMF ships and fewer errors in interpreting data.
PQFSBUJPOTXJUIMPXWBSJBUJPO t5IF64'PPEBOE%SVH"ENJOJTUSBUJPOTQSPDFTT
t6OTUBCMFBOBMZUJDBMTZTUFNTTJHOBMTGSPNUIF WBMJEBUJPOHVJEBODFSFDPNNFOETTUBUJTUJDBMQSP
BOBMZUJDBMQSPDFTTBEEWBSJBUJPOUPQSPEVDUJPO DFTTDPOUSPMUFDIOJRVFTUPNFBTVSFBOEFWBMVBUF
data. QSPDFTTTUBCJMJUZBOEQSPDFTTDBQBCJMJUZ
[
ABOUT THE AUTHORS
For more Author John McConnell is owner and director at Wysowl Pty Ltd in Queensland, Australia. He may be
information, reached at john@wysowl.com.au. Brian K. Nunnally, Ph.D., is in charge of process validation
go to at Pfizer in Sanford, NC. He may be reached at brian.nunnally@pfizer.com. Bernard McGarvey,
gxpandjvt.com/bios Ph.D., is process modelling group leader, Process Engineering Center at Eli Lilly and Company in In-
dianapolis, IN. He may be reached at mcgarvey_bernard@lilly.com.

INTRODUCTION UPEPHPPETDJFODF XFPVHIUUPDSFBUFTUBCMF

5IJTBSUJDMFDPOUJOVFTEJTDVTTJPOJOJUJBUFEJOi#MBNF PQFSBUJPOT6OGPSUVOBUFMZ NVDIPGUIFJOEVTUSZ
UIF-BCPSBUPSZ6OEFSTUBOEJOH"OBMZUJDBM&SSPSw IBTZFUUPEJTDPWFSUIJTUSVUI MFUBMPOFVTFJUUP

5IBUBSUJDMFHFOFSBUFENPSFDPNNFOUBOEEJT UIFJSBEWBOUBHF5PJMMVTUSBUFUIFTJUVBUJPO UXP
DVTTJPOUIBOBOZPUIFSBSUJDMFQVCMJTIFEJOUIJTDPM DPOUSPMDIBSUTBSFTIPXOJO'JHVSF
5IFZ
VNO BOEJUTPPOCFDBNFDMFBSUIBUSFBEFSTSFRVJSFE TIPXUIFSFTVMUTPGBQMBOUUSJBMXIPTFPCKFD
NPSFEFUBJMBOEHVJEBODF"TUIJTBSUJDMFXBTCFJOH UJWFXBTUPESJWFWBSJBUJPOUPNJOJNVNMFWFMT
XSJUUFO POFPGUIFBVUIPSTWJTJUFEBMBSHFQIBSNB JOFWFSZUIJOH5IFTBNFQFPQMFVTJOHUIFTBNF
DFVUJDBMTTJUFQSPEVDJOHCJPMPHJDBMQSPEVDUT&BSMJFS UFDIOPMPHZNBEFUIFTBNFQSPEVDUGPSUIFQFSJPE
JOUIFZFBS BTMPXBOEMPOHUFSNVQXBSEESJGUJO of the chart. There is no change in the science
UIFMFWFMPGBOBMZUJDBMFSSPSIBECFFOEFNPOTUSBUFE JOWPMWFE8IBUDIBOHFEXBTPQFSBUJPOBMSBUIFS
*OBEEJUJPO JUXBTOPUFEUIBUBTJHOJGJDBOUESPQJO UIBODIFNJDBMPSCJPMPHJDBM8IBUDIBOHFEXBT
UIFBWFSBHFPGUIFQSPEVDUJPOEBUBXBTNBUDIFE UIBUFWFSZPOFJOWPMWFE GSPNUIFQMBOUNBOBHFS
XJUIBTJNJMBSESPQJOUIFBWFSBHFGPSMBCPSBUPSZ EPXO CFDBNFJOUPMFSBOUPGWBSJBUJPOJOBOZ
SFGFSFODFTUBOEBSET'VSUIFSTUVEJFTSFWFBMFEUIBU GPSN5SBJOJOHXBTDPOEVDUFE PQFSBUJPOBMEFGJ
BOBMZUJDBMFSSPSXBTMJLFMZJODSFBTJOHWBSJBUJPOJO OJUJPOTXFSFDSFBUFE BOENFUIPENBTUFSTXFSF
UIFGPSNVMBUJPOPGUIFGJOBMQSPEVDU BQQPJOUFEUPFOTVSFBMNPTUFYBDUQFSGPSNBODF
*UXBTDMFBSUIBUBOBMZUJDBMFSSPSXBTFYDFTTJWF CFUXFFOTIJGUBOECFUXFFOPQFSBUPSBOEBOB
BOEUIBUJUOFFEFEUPCFSFEVDFE5IFDFMMDPVOUGPS MZTUSFQFBUBCJMJUZ*OTUSVNFOUTXFSFUFTUFEBOE
SFGFSFODFTUBOEBSETNFUUIFEFTJSFENJOJNVNMFWFM DBMJCSBUFEUPFOTVSFFYDFMMFOUSFQMJDBUJPOBDSPTT
POMZBCPVUPGUIFUJNF"QSPKFDUUPSFEVDF JOTUSVNFOUT#BDUFSJBGSPNPOMZPOFXPSLJOHDFMM
BOBMZUJDBMFSSPSXBTJOJUJBUFE4JYXFFLTBGUFSUIJT CBOLXFSFVTFEJOGFSNFOUBUJPO5IFBJNXBT
QSPKFDUXBTJOUSPEVDFE SFNBSLBCMFSFTVMUTIBECFFO OFWFSDPODFSOFEXJUIBDDVSBDZGPSBOZDIBSBD
BDIJFWFE$FMMDPVOUNFUUIFTUBOEBSEPGUIF UFSJTUJD5IFBJNXBTBMXBZTUPDSFBUFNBYJNVN
UJNF BOEUIFTUBOEBSEEFWJBUJPOGPSUIJTDFMMDPVOU QSFDJTJPO UPDPORVFSWBSJBUJPO BOEUPDSFBUF
XBTMFTTUIBOIBMGPGUIBUXIJDIFYJTUFECFGPSFUIF SFQFBUBCJMJUZ/PXIFSFXBTUIJTEPOFCFUUFSUIBO
QSPKFDUDPNNFODFE"RVJFUSFWPMVUJPOJTUBLJOH JOUIFMBCPSBUPSZ
QMBDFJOUIJTBOBMZUJDBMTZTUFN"OBMZUJDBMFSSPS 5IFSFBSFUXPFMFNFOUTUIBUPVHIUUPCFLFQUJO
IBTCFFOTMBTIFE BOEUIFQSPKFDUJTGBSGSPNPWFS NJOEXIFOFYBNJOJOH'JHVSF'JSTU JUTIPVME
JOUSVUI JUIBTCBSFMZCFHVO*OUFSFTUJOHMZ OFBSMZ CFDMFBSUIBUOPUPOMZXBTUIFGBDUPSZ JOUIJTDBTF
BMMUIFJNQSPWFNFOUXPSLIBTCFFOEPOFCZUIF UIFGFSNFOUBUJPOTUFQGPSBCJPMPHJD
TVDDFTTGVM
UFDIOJDJBOT*OUIJTFYBNQMF UIFTDJFODFSFNBJOT JODPORVFSJOHWBSJBUJPO CVUBMTPTPUPPXBTUIF
unchanged. It is the conduct of operations that has BOBMZUJDBMMBCPSBUPSZJOWPMWFE5IFMBCPSBUPSZ
JNQSPWFE$FOUSBMUPUIJTBSUJDMFJTUIFOPUJPOUIBU NBOBHFSBOEUIFUFDIOJDJBOTJOWPMWFESFEVDFE
JGXFBSFUPEPHPPETDJFODF XFBSFXFMMTFSWFEUP BTTBZWBSJBUJPOCZKVTUBTTJHOJGJDBOUBQSPQPSUJPO
TUBSUCZDPORVFSJOHWBSJBUJPOJOPQFSBUJPOT BTEJEUIFQSPEVDUJPOQFPQMF5IJTNVTUCFUSVF
PUIFSXJTF UIFDIBOHFJOGBDUPSZQFSGPSNBODF
GOOD SCIENCE REQUIRES GOOD XPVMEOPUIBWFCFFOTPPCWJPVT4FDPOEMZ JG
OPERATIONS UIFJOTUSVNFOUGBJMVSFOPUFEJOUIFQ)DIBSUIBE
1IBSNBDFVUJDBMTDPNQBOJFTBSFEFTJHOFE CVJMU BOE PDDVSSFECFGPSFUIFUSJBM UIFSFJTFWFSZDIBODF
NBOBHFECZTDJFOUJTUT5IJTJTPOMZBTPOFNJHIUFYQFDU UIBUJUXPVMEIBWFHPOFVOOPUJDFE*UJTBYJPN
/FBSMZBMXBZT POFPGUIFNPTUJNQPSUBOUDSJUFSJBGPS BUJDUIBUBTXFSFEVDFWBSJBUJPOJOBOZQSPDFTT
QSPNPUJPOXJMMCFUFDIOJDBMTLJMMTBOEBCJMJUZ5IJT FWFSTNBMMFSTJHOBMTDBOCFEFUFDUFEUISPVHIUIF
SFTVMUTJOQIBSNBDFVUJDBMTCVTJOFTTFTIBWJOHBDVMUVSF SFEVDFECBDLHSPVOESBOEPNiOPJTFw
TUSPOHMZCJBTFEUPXBSETUFDIOJDBMFYDFMMFODFCPUIBU This is a critical understanding if we are to do
BCVTJOFTTVOJUBOEBUBOJOEJWJEVBMMFWFM5FDIOJDBM good science. Nowhere is this truer than in the ana
FYDFMMFODFJTBWFSZHPPEPCKFDUJWF)PXFWFS XIFO MZUJDBMXPSME5IFMPXFSUIFWBSJBUJPOJOBTTBZT UIF
TVDIDPNQBOJFTFODPVOUFSBQSPCMFN UIFOBUVSFPGUIF FBTJFSJUJTUPEFUFDUEJTUVSCBODFTJOUIFBOBMZUJDBM
CVTJOFTTBOEUIFQFPQMFXIPTUBGGUIFNJTUPBEESFTT QSPDFTTBOEUPDPSSFDUUIFNCFGPSFUIFZDBVTFEFWJB
UIFQSPCMFNGSPNBTDJFOUJGJDPSUFDIOJDBMQFSTQFDUJWF UJPOTPSPUIFSUSPVCMF5IFDVTUPNFSPGUIFMBCPSB
5IJTDBOCFBUFSSJCMFNJTUBLF FTQFDJBMMZJGUIFQSPDFTT UPSZBMTPCFOFGJUT5IFMPXFSUIFWBSJBUJPOJTJOUIF
VOEFSFYBNJOBUJPOJTOPUTUBUJTUJDBMMZTUBCMF BTTBZ UIFFBTJFSJUJTGPSQSPEVDUJPOQFPQMFUPEFUFDU
TJHOBMTJOUIFQSPEVDUJPOEBUB'JHVSFTIPXTB
Some Actual Examples DIBSUPGMBCPSBUPSZDPOUSPMT SFGFSFODFNBUFSJBM
JO
#FGPSFXFBTLUIFTDJFOUJTUTJOUIFEJTDPWFSZ BOPUIFSDPNQBOZ5IFQSPEVDUJPOQFPQMFCFMJFWFE
EFWFMPQNFOU QSPEVDUJPO PSBOBMZUJDBMBSFBT UIFBTTBZUPCFJOBDDVSBUFBOEXFSFEFNBOEJOH

Figure 1: Results of a plant trial to reduce variation.
NPSFSFQMJDBUFTJOBOBUUFNQUUPJNQSPWFBTTBZBDDV EFNPOTUSBCMF5IJTXBTOPUQPTTJCMFXJUIWBSJBUJPO
SBDZ5IFBOBMZUJDBMMBCPSBUPSZNBOBHFSEJTBHSFFE BUUIFMFWFMQSJPSUPUIFTUFBEZTUBUFUSJBM
TVHHFTUJOHUIFQSPCMFNXBTJOBTTBZWBSJBUJPO
SBUIFSUIBOJOBDDVSBDZ)FBTTJHOFEBTUBUJTUJDJBO WHAT SHOULD BE THE INITIAL AIM?
UPESJWFWBSJBUJPOJOUIFDPOEVDUPGPQFSBUJPOTUPB 4UBCJMJ[JOHUIFQSPDFTTBOESFEVDJOHWBSJBUJPO
NJOJNVN"HBJO BESBNBUJDEFDSFBTFJOBOBMZUJ TIPVMECFUIFJOJUJBMBJNGPSFWFSZBOBMZUJDBMQSP
DBMFSSPSJTPCTFSWFE"TCFGPSF UIFJNQSPWFNFOU DFTT*OQBSUJDVMBS WBSJBUJPOJOMBCPSBUPSZPQFSB
JTFOUJSFMZPQFSBUJPOBM BOEOFBSMZBMMUIFJNQSPWF UJPOT XIJDINBTLTUIFDBVTBMSFMBUJPOTIJQTGSPN
NFOUXPSLXBTEPOFCZUIFBOBMZTUT5IFFOUJSF UIFTDJFOUJTUT PVHIUUPCFBOFBSMZUBSHFU
QSPKFDUMBTUFEGPSBXFFL/PDIBOHFXBTNBEFUP )BWFBOZPGVTFWFSNFUBOZCPEZXPSLJOHJO
the science. QIBSNBDFVUJDBMTPSCJPMPHJDTXIPJTOPUJOUFSFTUFE
(PPETDJFODFSFRVJSFTHPPELOPXMFEHFBOEB JOWBSJBUJPO BOEJGQPTTJCMF SFEVDJOHJU &WFSZ
HPPEVOEFSTUBOEJOHPGUIBUXIJDIJTCFJOHJOWFT DIFNJTU CJPMPHJTU WJSPMPHJTU BOBMZTU NBOBHFS PS
UJHBUFE5IJTSFRVJSFTVOEFSTUBOEJOHPGDBVTBM PQFSBUPSXJUIXIPNXFIBWFEJTDVTTFEUIJTTVCKFDU
SFMBUJPOTIJQT5IFDIBSUTJO'JHVSFDPNFGSPNUIF IBTCFFOJOBHSFFNFOUUIBUSFEVDJOHWBSJBUJPOJTB
TBNFUSJBMBTUIPTFSFTVMUTTIPXOJO'JHVSF5IF HPPEUIJOHUPEP4PNFNJHIUDMBJNUIBUJUJTTPNF
UXPWBSJBCMFTTIPVMEIBWFTIPXOBTUSPOHDPSSFMB UJNFTOPUQPTTJCMFJODFSUBJODJSDVNTUBODFTUIBU
UJPOCBTFEPOUIFTDJFODF CVUVOUJMUIFPQFSBUJPOT XFIBWFIJUUIFMJNJUTPGPVSUFDIOPMPHZ0UIFST
XFSFTUBCJMJ[FEXJUINJOJNVNWBSJBUJPO UIFTDJFO NJHIUCFBEBNBOUUIBUXIJMTUJUJTOFDFTTBSZUP
tists could not understand the process well enough SFEVDFWBSJBUJPO UIFJSIBOETBSFUJFECFDBVTFUIF
UPEPHPPETDJFODF*O'JHVSF OPUPOMZEPXFOPUF SFBMDBVTFTPGWBSJBUJPOMJFJOBEJGGFSFOUEFQBSU
BNVDISFEVDFETDBUUFSBOEBWBTUMZJODSFBTFE3 fac NFOU BOETPPO/FWFSUIFMFTT JUTFFNTXFBSFBMMJO
UPS CVUBMTPUIBUUIFBOHMFPGTMPQFPGUIFSFHSFTTJPO BHSFFNFOUUIBUSFEVDJOHWBSJBUJPOJTBHPPEUIJOHUP
MJOFJTGVOEBNFOUBMMZBMUFSFE UIFTIBMMPXTMPQF do. The reasons that understanding and reducing
JOUIFMFGUDIBSUPG'JHVSFJTDBVTFECZJOTUBCJM WBSJBUJPOJTBMXBZTBHPPEUIJOHUPEPBSFNBOZ
JUZ
'SPNBTDJFOUJTUTQFSTQFDUJWF CPUIBSFDSJUJDBM
VOEFSTUBOEJOHT"GUFSUIFUSJBM UIFEBUBNBEFTFOTF Polio Vaccine Clinical Supplies
and the correlation that always existed was clearly 'SPNBRVBMJUZQFSTQFDUJWF MPXFSWBSJBUJPONFBOT

Figure 2: Reduced variation in analytical error.
NPSFQSFEJDUBCMFBOECFUUFSRVBMJUZQSPEVDU+POBT BOEFWFOXIFOOPPSGFXEFGFDUJWFTBSFQSPEVDFE
4BMLVOEFSTUPPEXFMMUIFOFFEGPSSFQFBUBCJMJUZBOE 3FEVDUJPOJOWBSJBUJPONFBOTHSFBUFSVOJGPSNJUZBOE
QSFEJDUBCMFPVUDPNFTBTBLFZRVBMJUZJOEJDBUPS*O EFQFOEBCJMJUZPGQSPEVDU HSFBUFSPVUQVUQFSIPVS
UIFGJSTUCBUDIFTPGQPMJPWBDDJOFTXFSFNBOV HSFBUFSPVUQVUQFSVOJUPGSBXNBUFSJBM BOECFUUFS
GBDUVSFEGPSUIFNBTTJWFDMJOJDBMUSJBM0WFS DPNQFUJUJWFQPTJUJPO
w
EPTFTXFSFBENJOJTUFSFEXJUIPVUBOZTFSJPVTJODJ 6OGPSUVOBUFMZ ZFBSTMBUFSXFBSFTUJMMMFBSOJOH
EFOUTPSOFHBUJWFFGGFDUT5IF/BUJPOBM'PVOEBUJPO UIBU4IFXIBSUBOE%FNJOHXFSFDPSSFDU5PUIJTEBZ
GPS*OGBOUJMF1BSBMZTJTIBEEFNBOEFEUIBUUPIBWF XIFOBOBMZUJDBMNBOBHFSTJOUIFQIBSNBDFVUJDBMT
UIFJSWBDDJOFBDDFQUFEGPSUIFUSJBM NBOVGBDUVSFST JOEVTUSZBSFNBEFBXBSFUIBUUIFJSQSPDFTTFTDPOUBJO
XFSFSFRVJSFEUPNBLFTVDDFTTJWFCBUDIFT BMMPG VOOFDFTTBSZWBSJBUJPONBOZSFTQPOEXJUI iCVU*BN
XIJDIEFNPOTUSBUFEUIBUUIFMJWFWJSVTXBTDPN DPNQMJBOUXIBUJTUIFQSPCMFN w
QMFUFMZJOBDUJWBUFE0OMZUXPNBOVGBDUVSFSTNFUUIJT
DSJUFSJPOBOEPOMZUIFTFUXPQSPWJEFEWBDDJOFGPSUIF Why No Progress?
DMJOJDBMUSJBM"GUFSUIFTVDDFTTGVMUSJBM UIFGFEFSBM 4PNFBSFUSZJOHUPDPOWJODFUIFJOEVTUSZUIBU
HPWFSONFOUBTTVNFEPWFSTJHIUPGNBOVGBDUVSJOH UIFBQQSPBDIFTEFWFMPQFECZ4IFXIBSU %FNJOH
BOEMBSHFTDBMFWBDDJOBUJPO5IFSFRVJSFNFOUUPNBLF 4NJUI +VSBO )BSSZ BOEPUIFSTIPMETUIFQSPNJTF
TVDDFTTJWFJOBDUJWBUFECBUDIFTXBTESPQQFE4PPO PGJNQSPWFERVBMJUZBOEQSPEVDUJWJUZBTXFMMBT
BGUFSXBSET BNBONBEFQPMJPFQJEFNJDGPMMPXFE GFXFSEFWJBUJPOTBOESFHVMBUPSZJTTVFT
6OGPSUV
UIBUXBTDSFBUFEBMNPTUFYDMVTJWFMZCZBTJOHMFNBOV OBUFMZ DIBOHFJTPDDVSSJOHTMPXMZ*OUIFDBTFTUVEZ
facturer who was not part of the initial trial and who EFQJDUFEJO'JHVSF UIFMBCPSBUPSZNBOBHFSBOE
IBEOFWFSNBEFNPSFUIBOGPVSCBUDIFTJOBSPX TUBUJTUJDJBOXIPMFEUIJTBOBMZUJDBMSFWPMVUJPOQSF
XJUIPVUEFUFDUJOHMJWFQPMJPWJSVTJOGJOJTIFECBUDIFT TFOUFEUIFSFTVMUTPGUIFJSQSPKFDUUPDPMMFBHVFTBOE
-JWFWJSVTXBT JOTPNFCBUDIFT CFJOHNJTTFEEVSJOH QFFST5IFZJOUFOEFEUPFYQMBJOUIFNFUIPEPMPHZ
UFTUJOHBOEUIFTFCBUDIFTXFSFQBSBMZ[JOHBOELJMMJOH BOEEFNPOTUSBUFJUTCFOFGJUT
DIJMESFO0UIFSJTTVFTEJEFYJTU)PXFWFS UIFTVCTF 'PSUIFNPTUQBSU UIFJSBVEJFODFXBTVOSFTQPO
RVFOUJOWFTUJHBUJPOTIPXFEUIBUJGUIFSFRVJSFNFOU TJWF5IFZDPVMEOPUTFFBQSPCMFN(FOFSBMMZTQFBL
GPSSFQFBUBCJMJUZBOEQSFEJDUBCJMJUZIBECFFONBJO JOH UIFZXFSFNFFUJOHUIFSFRVJSFETUBOEBSET&WFO
UBJOFE UIFNBONBEFFQJEFNJDXPVMEOFWFSIBWF JGBTJNJMBSQSPKFDUJOUIFJSMBCPSBUPSJFTNJHIUZJFME
PDDVSSFECFDBVTFUIFQSPCMFNBUJDWBDDJOFXPVME TJNJMBSSFTVMUT XIZTIPVMEUIFZCPUIFSUPESJWF
OFWFSIBWFCFFOSFMFBTFEGPSVTF
BOBMZUJDBMFSSPSUPFWFOMPXFSMFWFMT /PBSHVNFOU
NPWFEUIFEFUSBDUPST/FJUIFSJNQSPWFETFSWJDFUP
Shewhart and Deming DVTUPNFSEFQBSUNFOUTOPSUIFQPUFOUJBMUPSFEVDF
&JHIUZZFBSTBHP %S%FNJOHFEJUFE%S4IFXIBSUT SFHVMBUPSZEFWJBUJPOTJNQSFTTFEUIFNOPSEJEUIF
TFNJOBMXPSL Economic Control of Quality of Manu- PQQPSUVOJUZUPQSPWJEFBCFUUFSQMBUGPSNGPSTDJFOUJGJD
factured Product
6OUJMIJTEFBUIJO %FNJOH XPSL OPXBOEJOUIFGVUVSF
QMFBEFEXJUIXFTUFSOCVTJOFTTUPXPSLBUVOEFS 6OUJMTFOJPSNBOBHFNFOUSFNPWFTPQUJPOTUPDPO
TUBOEJOHBOESFEVDJOHWBSJBUJPOJOFWFSZUIJOHUIFZ RVFSJOHWBSJBUJPOCZNBLJOHJUBTUSBUFHJDJNQFSBUJWF
EP%FNJOHTUBUFEi*UJTHPPENBOBHFNFOUUPSFEVDF XFPVHIUOPUUPCFTVSQSJTFEJGTPNFSFGVTFUPTXJUDI
UIFWBSJBUJPOPGBOZRVBMJUZDIBSBDUFSJTUJDXIFUIFS UIFJSGPDVTGSPNUFDIOJDBMUPPQFSBUJPOBMJTTVFT
UIJTDIBSBDUFSJTUJDCFJOBTUBUFPGDPOUSPMPSOPU 8IFOUSPVCMFPDDVSTJOUIFQSPDFTT UIFSFJTBTUSPOH

Figure 3: Before and after results of a steady-state trial (sst).
UFOEFODZGPSTDJFOUJTUTUPTFBSDIGPSUIFiTNPLJOH TVDINVTJOHT8IBUXFDBOEPJTUPXPSLOPXBOE
HVOw4PNFUJNFTJUFYJTUT BOETPNFUJNFTJUEPFTOPU GPSFWFSUPNJOJNJ[FUIFWBSJBUJPOJOPQFSBUJPOTUP
8IFSFJUEPFTFYJTU JUXJMMCFNVDIFBTJFSUPGJOEJOB HJWFUIFTDJFOUJTUTUIFCFTUDIBODFBUEPJOHHPPE
MPXWBSJBUJPOFOWJSPONFOU*ONBOZDBTFT IPXFWFS science.
XIBUFYJTUTJTOPUTPNVDIBTNPLJOHHVOBTB
UJOZGJSFDSBDLFSTBQMFUIPSBPGPQFSBUJPOBMJTTVFT THE FIRST PRINCIPLE
UIBUDPNCJOFUPQSPEVDFBOPJTZFOWJSPONFOUXJUI i4UBCJMJ[FGJSTUwJTUIFGJSTUQSJODJQMF
'JHVSF
IJHIWBSJBUJPOJOXIJDIJUJTWFSZEJGGJDVMUUPEPHPPE TIPXTBTUBCMFBOEBOVOTUBCMFQSPDFTT TJEFCZTJEF
science. BTBTFSJFTPGEJTUSJCVUJPOT
8IBUBSFUIFJNQMJDB
'SPNBDPNQMJBODFQFSTQFDUJWF SFEVDFEWBSJB UJPOTPGJOTUBCJMJUZ 'JSTU CZEFGJOJUJPO BOVOTUBCMF
UJPONFBOTGFXFSEFWJBUJPOT GFXFSSFTPVSDFTUJFE QSPDFTTJTOPUQSFEJDUBCMF"NPEFSO+POBT4BML
VQDPOEVDUJOHJOWFTUJHBUJPOTBOEQSFQBSJOHSFQPSUT XPVMESJHIUMZFYDMVEFUIFVOTUBCMF VOQSFEJDUBCMF

NPSFSFTPVSDFTEFEJDBUFEUPEPJOHUIFDPSFXPSL TVQQMJFSPGQSPEVDUPSPGBOBMZUJDBMTFSWJDFT*OBEEJ
5IJTFOBCMFTTFDVSJUZGPSBMM JF UIFDPNQBOZ UJPO VOUJMJUJTTUBCMF BQSPDFTTIBTOPLOPXODBQB
UIF64'PPEBOE%SVH"ENJOJTUSBUJPO BOEUIF CJMJUZ
0OFDBOEPUIFDBMDVMBUJPOT CVUUIFSFTVM
DPOTVNFS
UIBUTQSJOHTGSPNBQSFEJDUBCMF SFQFBU UBOUTPGUIFTFDBMDVMBUJPOTNFBOOPUIJOHJGUIFEBUB
BCMF BOEQSFDJTFBOBMZUJDBMQSPDFTTXJUIBLOPXO BSFVOTUBCMF8IBUEPFTUIJTJNQMZXIFOUIFMBCPSB
DBQBCJMJUZ UPSZDPOUSPMTJOWFTUJHBUFECZUIFBVUIPSTIBWFOFWFS
'SPNBOPQFSBUJPOBMQFSTQFDUJWF -JUUMFT-BX PODFCFFOTUBCMFBUUIFDPNNFODFNFOUPGJOWFTUJHB
FYQMBJOTXIZ%FNJOHXBTSJHIUXIFOIFDMBJNFE UJPOT 'JSTUBOEGPSFNPTU JOTUBCJMJUZNBLFTBNPDL
UIBUSFEVDJOHWBSJBUJPOJODSFBTFEPVUQVU*ODSFBTFE FSZPGUIFFTUJNBUFTQSPWJEFEGPSBOBMZUJDBMFSSPS
PVUQVUGSPNUIFTBNFSFTPVSDFT QFPQMFBOEFRVJQ *GUIFMBCPSBUPSZDPOUSPMTBSFVOTUBCMF OPEFHSFF
NFOU
NFBOTMPXFSDPTUT PGDPOGJEFODFDBOCFBQQMJFEUPUIFEFHSFFPGMJLFMZ
'JOBMMZ GSPNBTDJFOUJGJDQFSTQFDUJWF PQFSBUJOH BOBMZUJDBMFSSPSJOUIFGVUVSF XIJDIJTXIBUQSPDFTT
JOBMPXWBSJBUJPOFOWJSPONFOUSFTVMUTJOFBTJFS DBQBCJMJUZNFBTVSFTJOBMBCPSBUPSZ#FDBVTFQSPDFTT
detection of causal relationships and fewer errors DBQBCJMJUZJNQMJFTQSFEJDUJPO XIBUXJMMCFUIFMJLFMZ
JOJOUFSQSFUJOHEBUB$POTJEFSQSFDMJOJDBMUSJBMT BOBMZUJDBMFSSPSOFYUNPOUI
BHMBODFBU'JHVSF
*GUIFTDJFOUJTUTBSFPQFSBUJOHJOBMPXWBSJBUJPO TPPOSFWFBMTUIBUBOZNFBTVSFPGQSPDFTTDBQBCJMJUZ
FOWJSPONFOU UIFSFXJMMCFGFXFSUZQF*BOEUZQF** POMZIBTNFBOJOHJGUIFEBUBBSFSFBTPOBCMZTUBCMF
FSSPST
"UZQF*FSSPSPDDVSTJGXFDPODMVEFUIBU
'JOBMMZ IPXEPTDJFOUJTUTFTUBCMJTIDBVTBMSFMB
UXPDBOEJEBUFNPMFDVMFTQSPEVDFEEJGGFSFOUFGGFDUT UJPOTIJQTXIFOUIFEBUBBSFVOTUBCMF 5IF8JOUFS
XIFOJOGBDUUIFSFXBTOPEJGGFSFODFCFUXFFOUIFN JTTVFPGUIFJournal of Validation Technology

"UZQF**FSSPSPDDVSTJGJUJTDPODMVEFEUIBUUIFUXP JMMVTUSBUFTUIJTJTTVF*OPOFFYBNQMF JUSFTVMUFEJOB
DBOEJEBUFNPMFDVMFTQSPEVDFEUIFTBNFFGGFDU QPUFOUJBMSPPUDBVTFCFJOHNPWFEGSPNUIFCPUUPN
when in fact there exists a real difference in perfor PGUIFMJTUUPUIFUPQ5PPPGUFO TJHOJGJDBOUFSSPSTJO
NBODF0OFJTJODMJOFEUPXPOEFSIPXPGUFOIJHI JOUFSQSFUJOHUIFTDJFODFBSFNBEF*UJTOPUQPTTJCMF
MFWFMTPGBOBMZUJDBMFSSPSIBWFTFOUUIFXSPOHDBOEJ UPEPHPPETDJFODFXIFOUIFEBUBBSFTPVOTUBCMF
EBUFNPMFDVMFUPUIFDMJOJDBOEXIBUUIFBTTPDJBUFE )PXFWFS JGUIFEBUBFYIJCJUTUBCJMJUZ UIFZBSF
DPTUTNJHIUCF8FDBOOFWFSLOPXUIFBOTXFSUP QSFEJDUBCMF5IJTNBLFTUIFBOBMZUJDBMQSPDFTTUSVTU

Figure 4: The first principle; stabilize first.
XPSUIZBOEFBTJFSUPNBOBHF*UHSFBUMZTJNQMJGJFT QPJOUTSFWFBMBDIBOHFJOUIFTZTUFN USJHHFSJOHB

TDIFEVMJOHBOEBMMPXTVTUPQSPWJEFBOBMZUJDBMDBQB TFBSDIGPSSPPUDBVTFTXIJMFXIBUFWFSDIBOHFEJTTUJMM
CJMJUZBOETFSWJDFHVBSBOUFFTUIBUBDUVBMMZNFBO UIFSFUPCFGPVOE"MUFSOBUFMZ JGBEFMJCFSBUFDIBOHF
TPNFUIJOH4UBCMFEBUBSFWFBMDBVTBMSFMBUJPOTIJQT IBTCFFONBEF PGUFOWFSZGFXQPJOUTBSFOFFEFEUP
NVDINPSFSFBEJMZ4UBCJMJ[JOHBQSPDFTTJTBLJOUP EFNPOTUSBUFBOJNQSPWFNFOUUPUIFQSPDFTT"XFMM
lifting a fog that hitherto had concealed the truth DPOTUSVDUFEDPOUSPMDIBSUMFBETUPGBTUFS NPSFFGGFD
GSPNBMM5IJTBMMPXTUIFTDJFOUJTUTUPEPHPPETDJ UJWFJOUFSQSFUBUJPOPGUJNFTFSJFTEBUB-BCPSBUPSZ
FODFGBSNPSFPGUFO'FXFSUZQF*BOEUZQF**FSSPST controls are a good place to start.
BSFNBEF*OUIFMBCPSBUPSZ BOBMZUJDBMFSSPSDBOCF &WFSZFYBNQMFJOUIJTBSUJDMFDBNFGSPNCJP
FWFOGVSUIFSSFEVDFE*OQSPEVDUJPO ZJFMETSJTFBOE MPHJDBMQSPDFTTFT4PNFXFSFWBDDJOFTPUIFSTXFSF
DPTUTGBMM*OEJTDPWFSZBOEEFWFMPQNFOU TDJFOUJTUT CJPMPHJDBMUIFSBQFVUJDTCVUBMMXFSFCJPMPHJDBM
BSFBCMFUPEFUFDUTNBMMFSDIBOHFTJOUIFQFSGPS 0QFSBUJPOBMFYDFMMFODF JF SFEVDFEPQFSBUJPOBM
NBODFPGBNPMFDVMFPSDFMMBOEUPEPBCFUUFSKPCPG WBSJBUJPO
JTNPTUJNQPSUBOUXIFOUIFQPUFOUJBMGPS
TFMFDUJOHUIFNPTUQSPNJTJOHDBOEJEBUFUPTFOEUP WBSJBCJMJUZJOUIFTDJFODFJTIJHIFSBOEXIFOEBUB
the clinic. BSFTDBSDFPSFYQFOTJWF5IFSFGPSF JOCJPMPHJDBM
BOBMZUJDBMQSPDFTTFTUIFOFFEUPBDIJFWFPQFSBUJPOBM
BIOLOGICAL ASSAYS FYDFMMFODFJTHSFBUFSUIBONJHIUVTVBMMZCFUIFDBTF
#ZUIFJSOBUVSF CJPMPHJDBMBTTBZTBSFVTVBMMZNPSF 5IFTBNFDBOCFTBJEPGEFWFMPQNFOUBSFBTXIFSF
WBSJBCMFUIBOUIFJSDIFNJDBMDPVOUFSQBSUT*UJTUPP EBUBBSFNVDINPSFTDBSDF*GXFDPNCJOFUIFTFUXP
FBTZUPTIFMUFSCFIJOEXIBUTFFNTUPCFVOBWPJE DPOTJEFSBUJPOT JUJTEJGGJDVMUUPBWPJEUIFDPODMVTJPO
BCMFWBSJBUJPO BOEUPDMBJNUIBUUIFMFWFMPGWBSJB UIBUBTTBZEFWFMPQNFOUGPSCJPMPHJDTJTPOFLFZBSFB
UJPOPCTFSWFEJTJOIFSJUJOUIFCJPMPHZBOEMBSHFMZ XIFSFUIFSFRVJSFNFOUUPEFTJHOGPSPQFSBUJPOBM
VOBWPJEBCMF*OBOBUUFNQUUPPWFSDPNFUIJTIJHI FYDFMMFODFBOESPCVTUOFTTJTBUJUTHSFBUFTU
MFWFMPGWBSJBUJPO BDPNNPOSFBDUJPOJTUPBEENPSF
SFQMJDBUFTBOENPSFDPTU)PXFWFS JGBDPOUSPMDIBSU CONCLUSION
NBEFXJUIMBCPSBUPSZSFGFSFODFTUBOEBSETTIPXT *OQBSU UIF'%"Guidance for Industry-Process Valida-
JOTUBCJMJUZ UIFJOFWJUBCMFDPODMVTJPOJTUIBUUIFTBNF tion: General Principles and Practices
TUBUFT
QFPQMFBOEJOTUSVNFOUTDBOQSPEVDFSFTVMUTXJUI We recommend that a statistician or person
SFEVDFEWBSJBUJPOJGPOMZUIFZDPVMETUBCJMJ[FUIFQSP with adequate training in statistical process control
DFTT'JHVSFTIPXTUXPSFDFOUFYBNQMFTPGSFGFSFODF techniques develop the data collection plan and sta-
TUBOEBSEQFSGPSNBODFJOCJPMPHJDBMBTTBZT#PUIBSF tistical methods and procedures used in measuring
VOTUBCMF5IJTNFBOTUIBUTUBCJMJ[JOHUIFQSPDFTTXJMM and evaluating process stability and process capabil-
TJHOJGJDBOUMZSFEVDFBOBMZUJDBMFSSPS3FHBSEMFTTPG ity. Procedures should describe how trending and
XIFUIFSUIFBTTBZVOEFSFYBNJOBUJPOJTDIFNJDBMPS calculations are to be performed and should guard
CJPMPHJDBMJOOBUVSF TUBCJMJUZJTNPSFPGUFOBOPQFSB against overreaction to individual events as well as
UJPOBMJTTVFUIBOJUJTBUFDIOJDBMJTTVF8IFOJUJTB against failure to detect unintended process variabil-
UFDIOJDBMJTTVF DBVTFTGPSUIFUSPVCMFDBOCFGPVOE ity. Production data should be collected to evaluate
NVDINPSFSBQJEMZBOEXJUINPSFDFSUBJOUZXIFO process stability and capability. The quality unit
UIFBTTBZJTTUBCMFXJUINJOJNVNWBSJBUJPO$PO should review this information. If properly carried
TJEFSUIFDIBSUTBU'JHVSF0ODFUIFDPSSFDUDPOUSPM out, these efforts can identify variability in the pro-
CBOEIBTCFFODBMDVMBUFE PGUFOPOMZPOFUPUISFF cess and/or signal potential process improvements.

Figure 5: Unstable laboratory controls (for biologics).
5IFSFJTMJUUMFSPPNGPSJOUFSQSFUBUJPOPGUIJT REFERENCES
TUBUFNFOU'%"JTEFNBOEJOHTUBCJMJUZBTB +.D$POOFMM #/VOOBMMZBOE#.D(BSWFZ i#MBNFUIF
NJOJNVNTUBOEBSE BOEXJUIHPPESFBTPO*OUIF -BCPSBUPSZ6OEFSTUBOEJOH"OBMZUJDBM&SSPS wJournal of
NBKPSJUZPGDBTFT JOTUBCJMJUZJTDBVTFECZPQFSB Validation Technology 7PMVNF /VNCFS 4VNNFS
UJPOBMSBUIFSUIBOUFDIOJDBMBTQFDUT)PXFWFS #,/VOOBMMZBOE+4.D$POOFMM Six Sigma in the
TDJFOUJTUTUFOEUPFYBNJOFBOZJTTVFGSPNBUFDIOJ Pharmaceutical Industry $3$1SFTT
DBMPSTDJFOUJGJDQFSTQFDUJWFCFDBVTFUIBUJTIPX 1"0GGJU The Cutter Incident :BMF6OJWFSTJUZ1SFTT
UIFZBSFUSBJOFEBOECFDBVTFUIFDVMUVSFPGNPTU
QIBSNBDFVUJDBMCVTJOFTTFTIBTBTUSPOHUFDIOJDBMPS 8"4IFXIBSU Economic Control of Quality of Manufac-
TDJFOUJGJDCJBT5IJTOFFEOPUCFBOJTTVF QSPWJEJOH tured Product 7BO/PTUSBOE
we understand that trying to do good science in an 8&EXBSET%FNJOH i0O4PNF4UBUJTUJDBM"JET5PXBSE&DP
VOTUBCMFQSPDFTTWBSJFTGSPNEJGGJDVMUUPJNQPT OPNJD1SPEVDUJPO wInterfaces 7PM /P "VHVTU
TJCMF QSPWJEJOHXFVOEFSTUBOEUIBUiTUBCJMJ[FGJSTUw +.D$POOFMM #/VOOBMMZBOE#.D(BSWFZ 5IF%PT
PVHIUUPCFUIFGJSTUQSJODJQMF BOEQSPWJEJOHUIBU "OE%POUTPG$POUSPM$IBSUJOH1BSU Journal of Vali-
XFVOEFSTUBOEUIBUOPXIFSFJTUIJTNPSFJNQPSUBOU dation Technology 7PMVNF /VNCFS 8JOUFS
UIBOJOUIFMBCPSBUPSZ '%" Guidance for Industry-Process Validation: General
Principles and Practices +BOVBSZ JVT

Meeting Specifications is Not

Good EnoughThe Taguchi Loss
Function
Analysis and Control of Variation is dedicated t5BHVDIJTDPODFQUTMFEUPUIFEFWFMPQNFOUPG

to revealing weaknesses in existing approaches to UIFNPTUDPNQMFUFEFGJOJUJPOPGRVBMJUZ"MM
understanding, reducing, and controlling variation DIBSBDUFSJTUJDTTIPVMEIBWFNJOJNVN TUBCMF
and to recommend alternatives that are not only WBSJBUJPOBSPVOEBOPQUJNVNWBMVF
based on sound science, but also demonstrably work.
The objective of the column is to combine sound sci-
ence with proven practical advice. INTRODUCTION
Reader comments, questions, and suggestions will %S8&EXBSET%FNJOH UIF"NFSJDBODSFEJUFE
help us fulfill our objective for this column. Please XJUIUIFRVBMJUZBOEQSPEVDUJWJUZSFWPMVUJPOJOQPTU
send your comments and suggestions to column XBS+BQBO XBTSFHVMBSMZRVPUFEBTTBZJOHUIBUJUXBT
coordinator John McConnell at john@wysowl.com. HPPENBOBHFNFOUUPSFEVDFUIFWBSJBUJPOJOBOZ
au or managing editor Susan Haigney at shaigney@ RVBMJUZDIBSBDUFSJTUJD XIFUIFSUIJTDIBSBDUFSJTUJD
advanstar.com. CFJOBTUBUFPGDPOUSPMPSOPU BOEFWFOJGGFXPS
OPEFGFDUJWFTPSPVUPGTQFDJDBUJPO 004
SFTVMUT
BSFPDDVSSJOH
5PNBOZXFTUFSOFST UIJTTFFNFE
KEY POINTS DISCUSSED JMMPHJDBM8IZSFEVDFWBSJBUJPOJGTQFDJDBUJPOTBSF
The following key points are discussed: BMSFBEZCFJOHNFU )PXFWFS BTIPSUTUSPMMUISPVHI
t(PPETDJFODFSFRVJSFTTUBCMFQSPDFTTPQFSBUJPOT UIFRVBMJUZBOENBOBHFNFOUMJUFSBUVSFPGUIFQBTU
XJUIMPXWBSJBUJPO ZFBSTTIPXTBDPNNPOUISFBE%FNJOHXBTOPU
t*UJTHPPENBOBHFNFOUUPSFEVDFUIFWBSJB UIFPOMZPOFUPTVHHFTUUIBUSFEVDJOHWBSJBUJPOXBT
UJPOJOBOZRVBMJUZDIBSBDUFSJTUJD XIFUIFSUIJT CFOFDJBM FWFOUIPVHITQFDJDBUJPOTXFSFCFJOH
DIBSBDUFSJTUJDCFJOBTUBUFPGDPOUSPMPSOPU BOE NFU%S+PIO-JUUMF BGUFSXIPN-JUUMFT-BXJT
FWFOJGGFXPSOPEFGFDUJWFTBSFCFJOHQSPEVDFE OBNFE QSPWFENBUIFNBUJDBMMZUIFMJOLTCFUXFFO
t5BHVDIJTNFUIPET %FNJOHTBQQSPBDI 4JY UISPVHIQVUWPMVNF XPSLJOQSPHSFTT BOEDZDMF
4JHNB BOE-JUUMFT-BXBMMEFNPOTUSBUFUIBU UJNF)FGVSUIFSEFNPOTUSBUFEUIBUDPORVFSJOH
SFEVDJOHWBSJBUJPOJTBOFDPOPNJDJNQFSBUJWF WBSJBUJPOJOQSPDFTTPXSFEVDFEDZDMFUJNF*O
t.BOZTDJFOUJTUTBOENBOBHFSTGJOEUIFNTFMWFT UVSO XPSLJOQSPHSFTTNVTUUIFOGBMM UISPVHIQVU
USBQQFEJOUIFPMEiHPBMQPTUwNFOUBMJUZGPS WPMVNFNVTUSJTF PSTPNFDPNCJOBUJPOPGCPUI
specifications NVTUPDDVS"DDPSEJOHUP-JUUMF SFEVDJOHWBSJBUJPO
t5IF5BHVDIJ-PTT'VODUJPOBQQSPBDIEFNPO QBZTPGGCFDBVTFJUQSPWJEFTBOPQQPSUVOJUZUP
TUSBUFTIPXSFEVDJOHWBSJBUJPOSFEVDFTDPTUT JODSFBTFUISPVHIQVUGSPNUIFTBNFNJYPGQFPQMF
FWFOJGBMMPVUDPNFTNFFUTQFDJGJDBUJPO BOENBDIJOFT

[
ABOUT THE AUTHORS
For more Author John McConnell is owner and director at Wysowl Pty Ltd in Queensland, Australia. He may be
information, reached at john@wysowl.com.au. Brian K. Nunnally, Ph.D., is in charge of process validation
go to at Pfizer in Sanford, NC. He may be reached at brian.nunnally@pfizer.com. Bernard McGarvey,
gxpandjvt.com/bios Ph.D., is process modelling group leader, Process Engineering Center at Eli Lilly and Company in In-
dianapolis, IN. He may be reached at mcgarvey_bernard@lilly.com.

Figure 1: The original industrial view of variation. IN THE BEGINNING...

LSL USL The old view states that A is a "UUIFUJNFPGUIFJOEVTUSJBMSFWPMVUJPOXIFOUIF
good result, and that B and C are STUBUUFNQUTUPNBLFJOUFSDIBOHFBCMFQBSUTXFSF
equally bad. VOEFSXBZ POFCSFBLUISPVHIXBTUIFJOWFOUJPO
However, in terms of their PGHPBOEOPHPHBVHFT'PSJOTUBODF UFTUJOHUIF
physical properties, B and C may EJBNFUFSPGBOBYMFNJHIUCFEPOFXJUIUXPSJOHTPG
All bad All good All bad be nearly identical.
TUFFM POFTMJHIUMZTNBMMFSUIBOUIFPUIFS5IFMBSHFS
Is there a better way to describe HPHBVHFNVTUUPWFSUIFBYMF UIFTNBMMFSOPHP
their varying degrees of
goodness or badness? HBVHFNVTUOPU5IJTCSFBLUISPVHINBEFQPTTJCMF
UIFNBOVGBDUVSFPGDPNQPOFOUTJOMPDBUJPOTGBS
A B C SFNPWFEGSPNUIFQPJOUPGBTTFNCMZBOEJTB
DPSOFSTUPOFPGNBTTQSPEVDUJPO4PPOBGUFSXBSET
Figure 2: The Taguchi Loss Function Curve. UIFDPODFQUPGTQFDJDBUJPOT PSUPMFSBODFT
XBT
EFWFMPQFE*UXBTBUUIJTQPJOUUIBUUIFSTU
Taguchi Loss The vertical height between EFOJUJPOPGRVBMJUZEFWFMPQFE
Function Curve the baseline and the curve All characteristics must remain within the go,
is the amount of loss. The
further one moves from the
no-go specifications.
Optimum Value, the greater 5IJTJTUIFXFMMLOPXOiHPBMQPTUwBQQSPBDIUP
is the loss. RVBMJUZ1SPWJEJOHBMMPVUDPNFTTUBZFECFUXFFOUIF
HPBMQPTUT UIFTQFDJDBUJPOT
BMMXBTXFMM0GUFO
QIBSNBDFVUJDBMPSHBOJ[BUJPOTCFMJFWFUIBUJGBSFTVMU
Very little loss.
NFFUTTQFDJDBUJPO OPUPOMZEPFTOPUIJOHNPSF
Greater loss. OFFEUPCFTBJEBCPVUJU CVUBMTPOPUIJOHNPSF
TIPVMECFTBJEBCPVUJU*HOPSJOHUIFWBSJBUJPOJOB
OV Total loss
LFZDIBSBDUFSJTUJDCFDBVTFJUNFFUTTQFDJDBUJPODBO
CFBUFSSJCMFCVUBMMUPPDPNNPONJTUBLF%FTQJUF
"UUIFDPSFPG4JY4JHNBJTBOVOEFSTUBOEJOH ZFBSTPGQSPHSFTT NBOZTDJFOUJTUTBOENBOBHFST
PGWBSJBUJPO SBUIFSUIBOJEFBTTVDIBT%."*$ OEUIFNTFMWFTUSBQQFEJOUIFPMEiHPBMQPTUw
EFOF NFBTVSF BOBMZ[F JNQSPWF DPOUSPM
UIF NFOUBMJUZ
TUSVDUVSFEJNQSPWFNFOUBQQSPBDI PSUIFiCFMUTw "GUFS4IFXIBSUDPOUSJCVUFEUIFOPUJPOTPG
TZTUFN5IFDFOUSBMJEFBCFIJOE4JY4JHNBJTUIBU TUBUJTUJDBMTUBCJMJUZBOEJOWFOUFEUIFDPOUSPMDIBSU
UIFWBSJBCJMJUZPGBOZJNQPSUBOUFWFOUJOUIFQSPDFTT UIFTFDPOEEFOJUJPOPGRVBMJUZXBTEFWFMPQFE
PVHIUUPCFOPNPSFUIBOIBMGUIBUBMMPXFECZ All characteristics must remain stable and within
UIFTQFDJDBUJPOT4UBUFEBOPUIFSXBZ UIFOFBSFTU the go, no-go specifications.
TQFDJDBUJPONVTUCFTJYTUBOEBSEEFWJBUJPOTBXBZ 5IJTXBTBTJHOJDBOUBEWBODF CVUCPUI%FNJOHT
GSPNUIFQSPDFTTNFBO4PNFUJNFCFGPSFUIF BQQSPBDIBOEUIF4JY4JHNBNFUIPEPMPHZBSF
4JY4JHNBDPODFQUXBTEFWFMPQFEJO.PUPSPMB TPNFXIBUBUPEETXJUIUIJTEFOJUJPO CFDBVTF
%S(FOJDIJ5BHVDIJ XJOOFSPGUIF%FNJOH CPUIBSFNPSFFYBDUJOH8JUIIJTMPTTGVODUJPO
1SJ[FGPS*OEJWJEVBMT DSFBUFEBDPODFQUXFDBMM DPODFQU 5BHVDIJXBTBCMFUPEFNPOTUSBUFUIBU
UIF5BHVDIJ-PTT'VODUJPO
5IF5BHVDIJ-PTT SFEVDJOHWBSJBUJPOCFMPXTQFDJDBUJPOTXBTUIFCFTU
'VODUJPOBHBJOEFNPOTUSBUFTUIBU%FNJOHXBT FDPOPNJDBMUFSOBUJWF4PXBTUIFDSFBUPSPGUIFDPSF
DPSSFDUUPDMBJNUIBUXFPVHIUUPDPOUJOVFUPSFEVDF 4JY4JHNBDPODFQUT #JMM4NJUI
WBSJBUJPOFWFOUIPVHIBMMTQFDJDBUJPOTXFSFCFJOH *OUIFT %S4IFXIBSUOPUFEUIBUUIFDPOUSPM
NFU PGRVBMJUZXBTFTTFOUJBMMZBOFDPOPNJDQSPCMFN
.BOZNBOBHFSTBOETDJFOUJTUTGPVOEB 5IFUJUMFPGIJTCPPL Economic Control of
SFRVJSFNFOUGPSTVDIEFNBOEJOHQSFDJTJPOEJGDVMU Quality of Manufactured Product NBLFTUIJTDMFBS
UPVOEFSTUBOE4PNFGPVOEJUJNQPTTJCMFUPCFMJFWF %S(FOJDIJ5BHVDIJGPMMPXFEBOEFOIBODFEUIJT
5IFNJOETFUJONPTUDPNQBOJFTJTTVDIUIBUJG MJOFPGUIJOLJOH)FEFOFEUIFDPTUPGQPPS
UIFDIBSBDUFSJTUJDNFFUTTQFDJDBUJPOPSTUBOEBSE RVBMJUZBTUIFUPUBMMPTTJODVSSFECZTPDJFUZEVFUP
OPGVSUIFSJNQSPWFNFOUJTOFDFTTBSZOPSJTJU WBSJBUJPOBOEQPPSRVBMJUZ5BHVDIJXBTQBTTJPOBUF
XBSSBOUFE*ONPTUDBTFT BOZSFTVMUPVUTJEFPGB BCPVURVBMJUZUPUIFQPJOUXIFSFIFDMBJNFEUIBU
TQFDJDBUJPOPSTUBOEBSEJTUSFBUFEBTBOBTTJHOBCMF UIFNBOVGBDUVSFSPGQPPSRVBMJUZ XJUIQBSUJDVMBS
PSTQFDJBMDBVTFPGWBSJBUJPO5IJT PGDPVSTF JTB SFGFSFODFUPSFXPSLBOESFKFDUT
XBTGBSXPSTFUIBO
NJTUBLF BUIJFG8IFOBUIJFGTUFBMTGSPNBOFJHICPS
IFIBTHBJOFEBOEIJTOFJHICPSIBTMPTU CVUUIFOFU
FDPOPNJDJNQBDUPOTPDJFUZJTOJM3FHBSEMFTTPG

XIPIPMETJU UIFXJMMTUJMMCFJOWFTUFEPSTQFOU
Figure 3: An asymmetrical Taguchi Loss Function Curve.
Taguchi Loss The vertical height between
POHPPETBOETFSWJDFT)PXFWFS JGBNBOVGBDUVSFS Function Curve the baseline and the curve
is the amount of loss. The
UISPXTBXBZJOSFKFDUTBOESFXPSL UIFDPTUPG further one moves from the
Optimum Value, the greater
XBTUFESFTPVSDFTDBOOFWFSCFSFDPWFSFECZFJUIFS is the loss.
UIFDPNQBOZPSCZTPDJFUZ*OBEEJUJPO BQSPDFTT In this case the loss
XJUIIJHIMFWFMTPGWBSJBUJPOJOQSPDFTTPXXJMM increases faster to the left
of the OV than it does to
IBWFMPXFSUISPVHIQVUUIBUXPVMECFUIFDBTFJGUIF the right.
WBSJBUJPOJOQSPDFTTPXXBTMPXFS5IFEJGGFSFODF OV
in throughput translates directly into unit costs. Copyright John McConnell 2011
)JHIFSWBSJBUJPOJOQSPDFTTPXDPTUTNPOFZ5IJT
MPTTJTQFSNBOFOU Figure 4: Finding the area of loss.
"VTFGVMXBZUPJOUSPEVDFUIF5BHVDIJ-PTT
'VODUJPOJTUPCFHJOXJUIUIFTVQQPTJUJPOUIBU
The area of loss
EFTJHOFOHJOFFST DIFNJTUT BOECJPMPHJTUTJOUFOTFMZ (shaded) is dictated
EJTMJLFWBSJBUJPO5IFZQSFGFSQFSGFDUQSFDJTJPOBOE by the steepness of
IBWFBOPQUJNVNWBMVFJONJOE)PXFWFS UIFTF the loss function
TDJFOUJTUTVOEFSTUBOEUIBUQFSGFDUJPODBOOPUCF curve.
BDIJFWFECZNFSFNPSUBMTBOEUIFNBDIJOFTUIFZ
NBLF BOETPUIFZDSFBUFTQFDJDBUJPOTPSUPMFSBODFT
UPEFMJOFBUFUIFCPVOEBSJFTPGQFSNJTTJCMF
WBSJBUJPO8IBUUIFEFTJHOFSTBSFUFMMJOHVTXIFO
UPMFSBODFTBSFTFUJTOPUIPXNVDIWBSJBUJPOJT
OV OV
0,UIFEFTJHOFSTBSFUFMMJOHVTXIBUUIFZEPOU
SFBMMZXBOUCVUBSFXJMMJOHUPUPMFSBUF3FHBSEMFTT Figure 5:3FEVDJOHWBSJBUJPOSFEVDFTUIFBSFBPGMPTT
PGUIJTUPMFSBODF UIFSFJTBMXBZTBOPQUJNVN
LSL USL LSL USL
WBMVF 07
'PSUIFTBLFPGTJNQMJDJUZBOEDMBSJUZ
UIPTFDIBSBDUFSJTUJDTXIFSFCJHHFSJTBMXBZTCFUUFS
JF IPSTFQPXFSPGBSPDLFUFOHJOF
BOETNBMMFS
JTBMXBZTCFUUFS JF XFJHIUPGBSPDLFU
XJMMCF
ignored.
THE OLD GOALPOSTS APPROACH

*NBHJOFXFDSFBUFQSPEVDU" XIPTFNFBTVSFE
DIBSBDUFSJTUJDGBMMTPOMZKVTUJOTJEFUIFVQQFS
TQFDJDBUJPO BTTFFOJO'JHVSF/PX JNBHJOF OV OV
XFNFBTVSFUIFOFYUQSPEVDU QSPEVDU# XIJDI
JTGPVOEUPCFPOMZKVTUPVUTJEFUIFTBNFVQQFS TAGUCHIS LOSS FUNCTION CONCEPT
TQFDJDBUJPO*OUFSNTPGUIFJSQIZTJDBMQSPQFSUJFT "TOPUFEJO'JHVSF 5BHVDIJFOWJTJPOFEBMPTT
BOEUIFJSQFSGPSNBODFJOUIFDVTUPNFSTIBOET GVODUJPODVSWF VTVBMMZ CVUOPUBMXBZT
BQBSBCPMB
UIFTFUXPQSPEVDUTBSFMJLFMZUPCF GPSBMMQSBDUJDBM UIFTJNQMFTUGPSNPGBMPTTGVODUJPODVSWF
XIJDI
JOUFOUT JEFOUJDBM*ONPTUDBTFT UIFTBNQMJOHBOE EFTDSJCFEUIFJODSFBTFJOMPTTPOBDPOUJOVVNBT
BOBMZUJDBMFSSPSJOWPMWFEXJMMCFHSFBUFSUIBOUIF PVUDPNFTNPWFEBXBZGSPNUIFPQUJNVNWBMVF
PCTFSWFEEJGGFSFODFCFUXFFOUIFTFUXPPVUDPNFT SBUIFSUIBOBTBHP OPHPTJUVBUJPO5IFWFSUJDBM
:FUUIFNFUIPEPMPHZVTFEJONPTUDPNQBOJFT IFJHIUGSPNUIFCBTFMJOF BUUIFCBTFMJOFUIFSFJTOP
will generally state that product B is outside MPTT
UPUIFMPTTGVODUJPODVSWFEFTDSJCFEIPXUIF
TQFDJDBUJPOTBOE UIFSFGPSF XBSSBOUTJOWFTUJHBUJPO BNPVOUPGMPTTJODSFBTFEBTSFTVMUTNPWFGVSUIFS
/PX MFUVTBEEPOFNPSFQSPEVDUUPUIFDIBSU BXBZGSPNUIFPQUJNVNWBMVF VOUJMFWFOUVBMMZ
QSPEVDU$ XIJDIJTXFMMPVUTJEFUIFTQFDJDBUJPO DPNQMFUFMPTTPDDVST

5IFPMEHP OPHPEFOJUJPOPGRVBMJUZTUBUFT 'PSUIPTFXJUIBCBTJDHSBTQPGUIFDPODFQUPG
UIBUBMMQSPEVDUTPVUPGTQFDJDBUJPOBSFFRVBMMZ WBSJBUJPO 5BHVDIJTBQQSPBDINBEFJOUVJUJWFTFOTF
CBE*OUIJTDBTF TVDIBOBQQSPBDIXPVMEMVNQ 4VQQPTFPOFJTVTJOHUIFHP OPHPBQQSPBDIBOE
UPHFUIFSPVUDPNFT#BOE$ EFTQJUFUIFJSCFJOHWFSZ UIFTQFDJDBUJPOTGPSQ)GPSBDFSUBJOQSPDFTT
different. QBSBNFUFSBSFUP"SFBEJOHPGXJMM
CFBDDFQUFEBTJOTQFDJDBUJPO CVUPOFPG
XJMMCFPVUPGTQFDJDBUJPO FWFOUIPVHIUIFSF

JTOPEJTDFSOBCMFDIBOHFJOUIFQSPEVDUPSJOJUT FMEGBJMVSFJTJHOPSFE JT

QFSGPSNBODF UIFTBNQMJOHBOEBOBMZUJDBMFSSPS
NBZXFMMCFHSFBUFSUIBOUIFEJGGFSFODFCFUXFFO k $PTUPGGBJMJOHTQFDJDBUJPO
these data). The Taguchi approach sees these data "WFSBHFTQFDJDBUJPOMJNJU

OPUTPMFMZPOUIFJSQPTJUJPOSFMBUJWFUPTQFDJDBUJPOT
CVUPOBDPOUJOVVN5IFUBSHFUPSPQUJNVNWBMVF
JT0VSMPTTJODSFBTFTDPOUJOVBMMZUIFGVSUIFS UNDERSTANDING
POFEFQBSUTGSPNUIJTPQUJNVN/PUFIPXBUUIF
07UIFMPTTJTNJOJNJ[FE CVUJTOPU[FSP"CTPMVUF LOSSES DUE TO VARIATION
MJNJUTBSFTUJMMOFDFTTBSZGPSSFMFBTFBOEDPNQMJBODF 5IFOFYUTUFQJTUPPWFSMBZUIFMPTTGVODUJPODVSWF
SFBTPOT CVUJGMPTTJTTFFOPOBDPOUJOVVN JUPQFOT POUPUIFSFMFWBOUEJTUSJCVUJPO5IFBSFBPGJOUFSTFDU
OFXQFSTQFDUJWFTBOEPQQPSUVOJUJFTUPVOEFSTUBOE PGUIFTFUXPDVSWFTSFQSFTFOUTUIFBNPVOUPGMPTT
BOEJNQSPWFUIFQSPDFTTBOEQSPEVDUIJUIFSUP XIJDIDBOCFDBMDVMBUFEGSPNFRVBUJPO4PNF
unnoticed. MPTTGVODUJPODVSWFTBSFMPXBOEBU BMMPXJOH
5IF5BHVDIJ-PTT'VODUJPODVSWFOFFEOPUCF DPOTJEFSBCMFNPWFNFOUBXBZGSPNUIFPQUJNVN
TZNNFUSJDBM BMUIPVHIJUPGUFOJT'JHVSFTIPXT WBMVFCFGPSFTJHOJDBOUMPTTJTJODVSSFE0UIFSTBSF
BOFYBNQMFXIFSFUIFDVSWFJTBTZNNFUSJDBM5IF UBMMBOETUFFQ XIFSFFWFOBTNBMMEFWJBUJPOGSPN
MPTTJODSFBTFTGBTUFSUPUIFSJHIUPGUIF07UIBOJU UIFPQUJNVNWBMVFDBOJODVSDPOTJEFSBCMFMPTT BT
EPFTUPUIFMFGU5IFEFUFSNJOBUJPOPGUIFTIBQFPG OPUFEJO'JHVSF5IFTUFFQOFTTPGUIJTDVSWFJT
UIFDVSWFJTUFDIOJDBMJOOBUVSF SBUIFSUIBOQVSFMZ ESJWFOCZUIFNBHOJUVEFPGUIFk factor.
statistical. *ONPTUDBTFT XPSLJOHPOUIFFBSMJFSTUFQTJOUIF
5IF%FNJOH 5BHVDIJ BOE4JY4JHNBBQQSPBDIFT QSPDFTTXJUIQBSUJDVMBSSFHBSEUPJOQVUTBOETFUVQ
EFNBOEBCBOEPONFOUPGSFMJBODFPOUIFHP OPHP PGUIFQSPDFTT QSPWJEFTTJHOJDBOUMFWFSBHFCFDBVTF
TQFDJDBUJPOT*UJTTUJMMOFDFTTBSZUPTFUBCTPMVUF UIFMPTTGVODUJPODVSWFTBSFOFBSMZBMXBZTTUFFQFS
MJNJUT CVUTVDIMJNJUTTIPVMEOPUCFUIFTPMF UIFSF5IFJNQPSUBODFPGXPSLJOHVQTUSFBNJOUIF
CBTJTGPSBTTFTTJOHUIFWBSJBUJPOBUBOZTUBHFPGUIF QSPDFTTDBOOPUCFPWFSTUBUFE5PPPGUFOBQMBOUT
QSPDFTT*OBEEJUJPO HP OPHPTQFDJDBUJPOTTIPVME NPTUUBMFOUFEQFPQMFDBOCFGPVOEXPSLJOHJOUIF
OFWFSCFVTFEBTUIFCBTJTGPSQSPDFTTDPOUSPM NJEEMFPSMBUUFSTUBHFTPGUIFQSPDFTT BGUFSWBSJBUJPO
5IF5BHVDIJMPTTGVODUJPOUBLFTTFWFSBMGPSNT JOUIFFBSMJFSFWFOUTIBTBMSFBEZEFTUBCJMJ[FEUIF
0OFRVBESBUJDGPSNPGDBMDVMBUJPOJTTIPXOJO process and shrouded cause and effect relationships.
FRVBUJPOTBOE
0UIFSTUBUJTUJDBMNFUIPETTVDIBT%0&XJMMIFMQUP
JTPMBUFUIPTFWBSJBCMFTUIBUIBWFUIFHSFBUFTUJNQBDU
PORVBMJUZBOEDPTUT"DPSOVDPQJBPGBEWBODFE
'PSBOZHJWFOQSPEVDU
TUBUJTUJDBMUFDIOJRVFTFYJTUT CVUJUJTCFZPOEUIF
TDPQFPGUIJTBSUJDMFUPDPWFSUIFN
Lossx = k(x-OV)2 <&RVBUJPO> 5BHVDIJTBQQSPBDIEFNPOTUSBUFTUXPJNQPSUBOU
JEFBT'JSTUMZ JUJTJODPODFSUXJUI%S%FNJOHT
8IFSF TUBUFNFOUUIBUTPNFMPTTXJMMCFJODVSSFEFWFOJG
BMMEBUBNFFUTQFDJDBUJPOTBOEUIBUCVTJOFTTFT
TIPVMEBMXBZTCFBUUFNQUJOHUPSFEVDFWBSJBUJPO
xWBMVFPGRVBMJUZDIBSBDUFSJTUJD /PUFIPXJO'JHVSF UIFTIBEFEBSFBCFUXFFO
OV0QUJNVN7BMVF UIFEJTUSJCVUJPOBOEUIFMPTTGVODUJPODVSWF
k "QSPQPSUJPOBMJUZDPOTUBOU BGVODUJPO EFNPOTUSBUFTUIBUMPTTFYJTUTFWFOUIPVHIBMM
of the failure cost structure) PVUDPNFTNFFUTQFDJDBUJPOT4FDPOEMZ JU
EFNPOTUSBUFTUIBUBTWBSJBUJPOJTSFEVDFE TPUPPJT
UIFBNPVOUPGMPTTJODVSSFE BTOPUFEJO'JHVSF*U
'PSBOZHJWFOQPQVMBUJPOPGQSPEVDUT POMZSFNBJOTUPJTPMBUFUIPTFWBSJBCMFTUIBUQSPWJEF
HSFBUFTUMFWFSBHF
Lossx = k(x2 + (Avex - OV )2) <&RVBUJPO> 5BHVDIJTDPODFQUTMFEUPUIFEFWFMPQNFOUPGUIF
NPTUDPNQMFUFEFOJUJPOPGRVBMJUZ BTGPMMPXT
All characteristics should have minimum, stable
"TJNQMJFEWFSTJPOPGUIFk factor
variation around an optimum value.
FYJTUTUIBUNBLFTDFSUBJOBTTVNQUJPOT %S5BHVDIJTDPODFQUJTPOFXBZUPJMMVTUSBUFIPX
JODMVEJOHUIBUUIFTQFDJDBUJPOTIBWF WBSJBUJPOJNQBDUTPOQSPDFTTFTBOEDPTUT4JY4JHNB
CFFOQSPQFSMZTFUBOEUIBUBOZDPTUPG JTBOPUIFS BTJT%FNJOHTBQQSPBDIBOE-JUUMFT-BX

8IBUJTDMFBSJTUIBUTFWFSBMEJGGFSFOUBQQSPBDIFTUP CONCLUSIONS
UIFTVCKFDUBSSJWFBUTUSJLJOHMZTJNJMBSDPODMVTJPOT 7BSJBUJPOJTOPUGSFFOPUFWFOXIFOBMMPVUDPNFT
as follows: BQQFBSUPNFFUTQFDJDBUJPOT7BSJBUJPODPOTVNFT
t7BSJBUJPODPTUTNPOFZ FWFOXIFOFWFSZUIJOH SFTPVSDFT NPOFZ BOEUJNF8IFSFBTJHOJDBOU
BQQFBSTUPNFFUTQFDJGJDBUJPOT QSPQPSUJPOPGUFDIOJDBMQFSTPOOFMTUJNFJT
t3FEVDJOHWBSJBUJPOBMXBZTSFEVDFTDPTUT FYQFOEFEJOWFTUJHBUJOH004SFTVMUT UJNFJTXBTUFE
t/PUBMMWBSJBCMFTIBWFBOFRVBMJNQBDUPODPTUT UIBUDPVMEIBWFCFFOTQFOUJNQSPWJOHUIFQSPDFTT
4PNFBSFWFSZTJHOJGJDBOUPUIFSTBSFOPU BOEUIFQSPEVDU3FEVDJOHWBSJBUJPOCZBDIJFWJOH
"DSJUJDBMQBSUPGUIF5BHVDIJBQQSPBDIJTUP BTUBCMFTUBUFBOEUIFONJOJNJTJOHWBSJBCJMJUZJT
EJTDPWFSXIJDIWBSJBCMFTIBSCPSUIFHSFBUFTU particularly crucial in highly regulated industries
potential to reduce costs. TVDIBTUIFQIBSNBDFVUJDBMBOEBFSPOBVUJDBM
-JUUMFT-BXFYQMBJOTIPXSFEVDFEWBSJBUJPOJO industries.
UIFPXPGNBUFSJBMUISPVHIBQSPDFTTDBOMFBE $MPTFFOPVHIJTOPUHPPEFOPVHIOPSJTCFJOH
UPJODSFBTFEUISPVHIQVUWPMVNF FTQFDJBMMZXIFO TBUJTFEXJUINFFUJOHTQFDJDBUJPOT
UIFFBSMJFTUFWFOUTJOUIFQSPDFTTBSFBEESFTTFE

5PFWFOUIFDBTVBMPCTFSWFS JUTPPOCFDPNFT
BQQBSFOUUIBUWBSJBCJMJUZJNQBDUTPOCPUIRVBMJUZ REFERENCES
BOERVBOUJUZ 8&%FNJOH On Some Statistical Aids toward Economic
"MUIPVHIUIFMPTTGVODUJPODVSWFDBOCF Production, Interfaces 7PM /P "VH
DBMDVMBUFEGPSBOZQSPDFTTWBSJBCMF JUJTOPU #,/VOOBMMZBOE+4.D$POOFMM Six Sigma in the
necessary to do this in order to understand the Pharmaceutical Industry $3$1SFTT
DFOUSBMJEFBTBOEUPDSFBUFQSPHSFTT'FXQFPQMF 84DIFSLFOCBDI The Deming Route to Quality and Pro-
VTJOHB5BHVDIJUZQFNJOETFUBDUVBMMZDPOEVDU ductivity $&&1SFTT
UIFDBMDVMBUJPOT)PXFWFS JUJTUIFDPODFQUUIBU +.D$POOFMM #/VOOBMMZBOE#.D(BSWFZ i)BWJOH*U
JTDSJUJDBM OPUUIFDBMDVMBUJPOT*GXFVOEFSTUBOE "MM wJournal of Validation Technology 7PMVNF /VNCFS
UIBUWBSJBUJPOBMXBZTBEETUPDPTUT BOEJGXFBSF 4QSJOH JVT
GPDVTFEPOSFEVDJOHWBSJBUJPOFWFSZXIFSFCVUXJUI
QBSUJDVMBSFNQIBTJTPOUIPTFWBSJBCMFTLOPXOUP
IBWFUIFHSFBUFTUJNQBDU TUFFQFTUMPTTGVODUJPO ARTICLE ACRONYM LISTING
DVSWFT
BEWBODFNFOUDBOCFNBEF*GJUJTHSBTQFE DMAIC Define, Measure, Analyze, Improve, Control
UIBUUIFFBSMZTUFQTJOUIFQSPDFTTHFOFSBMMZIBWF OOS Out of Specification
TUFFQFSMPTTGVODUJPODVSWFTGPSCPUIRVBMJUZBOE OV Optimum Value
RVBOUJUZ XPSLUPSFEVDFWBSJBUJPODBODPNNFODF
UIFSFJNNFEJBUFMZ5IFPXPOCFOFUTUISPVHIPVU
UIFQSPDFTTDBOCFESBNBUJD

Process Modeling:
A Powerful Weapon in the
Fight to Reduce VariationPart 1
ABSTRACT Q)SBOHF UP

8IJMFJUJTFBTZUPNBUIFNBUJDBMMZ
1SPDFTTNPEFMJOHNBLFTVTFPGGVOEBNFOUBMMBXTPG FYUSBQPMBUFUIFSFTVMUTPVUTJEFPGUIF%0&SBOHFT UIFSF
QIZTJDTUPBOBMZ[FUIFCFIBWJPSPGSFBMTZTUFNT1SPDFTT XPVMECFOPQIZTJDBMCBTJTGPSUIJT"OJOUFSBDUJPOUIBU
NPEFMJOHDBOIFMQJEFOUJGZOPUPOMZXIBUIBTPSXJMM XBTTNBMMPWFSUIF%0&SBOHFTNJHIUCFDPNFTJHOJGJDBOU
IBQQFOJOBTZTUFNCVUBMTPXIZJUIBTPSXJMMIBQQFO PVUTJEFUIF%0&SBOHFCVUUIFSFJTOPXBZUIFFYQFSJ
5IJTBQQSPBDIDBOCFVTFEGPSCPUITQFDJBMBOEDPNNPO NFOUFSDPVMELOPXUIJT
DBVTFWBSJBUJPOBOEJOBMMQIBTFTPGWBSJBCJMJUZSFEVDUJPO 5IFTFDPOEUZQFPGNPEFMJTCVJMUCBTFEPOUBLJOH
FGGPSUT6TFPGQSPDFTTNPEFMJOHUPSFEVDFWBSJBUJPOIBT JOUPBDDPVOULOPXOQIZTJDTSFMFWBOUUPUIFTZTUFNCFJOH
BQSPWFOUSBDLSFDPSEBOEXJMMCFVTFEFWFONPSFJOUIF TUVEJFE"TTDJFODFIBTQSPHSFTTFEPWFSIVOESFETPG
GVUVSFBTCPUITPGUXBSFBOEIBSEXBSFTZTUFNTHSPXJO ZFBST NBOZMBXTBOEGVOEBNFOUBMSFMBUJPOTIJQTIBWF
DBQBCJMJUZ5IJTQBQFSDPODFSOTUIFVTFPGQSPDFTTNPEFMT CFFOEFWFMPQFEUIBUDBOCFBQQMJFEBDSPTTNBOZTZTUFNT
DPOTJTUJOHPGNFDIBOJTUJDBOEGJSTUQSJODJQMFTNPEFMT including the following:
BOEOPUFNQJSJDBMNPEFMT"OZSFGFSFODFUPNPEFMTJO t$POTFSWBUJPOMBXTPGNBTT FOFSHZ NPNFOUVN
UIFSFTUPGUIJTBSUJDMFSFGFSUPUIFGPSNFSBOEOPUUIFMBU and charge
UFS5IJTQBQFSJTUIFGJSTUPGBUXPQBSUTFSJFT t5IFSNPEZOBNJDMBXTTVDIBTUIFJEFBMHBTMBXBOE
UIFTFDPOEMBXPGUIFSNPEZOBNJDT
INTRODUCTION t3FBDUJPOLJOFUJDT
1SPDFTTNPEFMJOHIBTTFFOXJEFBQQMJDBUJPOJODSFBUJOH t3BUFMBXTGPSNBTTBOEIFBUUSBOTGFS
WBMVFJOUIFQIBSNBDFVUJDBMJOEVTUSZCPUIJOQSPDFTT t&RVBUJPOTPGGMVJEEZOBNJDT UIF/BWJFS4UPLFT
EFWFMPQNFOUBOENBOVGBDUVSJOH
*OHFOFSBM FRVBUJPOT

NPEFMJOHDBOCFEJWJEFEJOUPUXPNBKPSDBUFHPSJFT 0OFPGUIFBEWBOUBHFTPGCVJMEJOHNPEFMTCBTFEPO
'JSTUMZ UIFSFBSFNPEFMTUIBUVTFMJUUMFPSOPLOPXMFEHF UIFTFSFMBUJPOTIJQTJTUIBUUIFSFJTOPXBCBTJTGPSFYUSBQ
PGUIFQIZTJDTJOWPMWFEJOUIFTZTUFNCFJOHTUVEJFE5IF PMBUJPOCFDBVTFJOGPSNBUJPOIBTCFFODPMMFDUFEPWFS
NPTUVUJMJ[FEGPSNPGUIFTFNPEFMTJOUIFQIBSNBDFVUJ NBOZZFBSTUPEFGJOFUIFSFHJPOTPGBQQMJDBCJMJUZPGUIFTF
DBMJOEVTUSJFTJTUIFFNQJSJDBMNPEFMUIBUBSJTFTGSPN GVOEBNFOUBMSFMBUJPOTIJQT5IFTFQIZTJDTCBTFENPEFMT
TUBUJTUJDBMBOBMZTJTBOEEFTJHOPGFYQFSJNFOUT %0&
DBOCFEJWJEFEJOUPUXPUZQFT

4VDINPEFMTBSFCZUIFJSOBUVSFNPTUBQQMJDBCMFGPS t.FDIBOJTUJDNPEFMT5IFTFBSFNPEFMTXIFSF
JOUFSQPMBUJOHCFUXFFOUIFJOQVUEBUBTQBDFPGUIFNPEFM QIZTJDBMSFMBUJPOTIJQTBSFCVJMUJOUPUIFNPEFMCVU
'PSFYBNQMF B%0&NJHIUCFDPOEVDUFEUPVOEFSTUBOE UIFNPEFMIBTVOLOPXOQBSBNFUFSTUIBUNVTUCF
UIFJNQBDUPGUFNQFSBUVSFBOEQ)POQVSJUZPGBSFBDUJPO FTUJNBUFECZSVOOJOHFYQFSJNFOUTPOUIFTZTUFN
QSPEVDUPWFSBUFNQFSBUVSFSBOHF $UP$
BOEB CFJOHTUVEJFE"OFYBNQMFPGUIJTJTBNPEFMUIBU
[
ABOUT THE AUTHORS
For more Author John McConnell is the coordinator of Analysis and Control of Variation and is the owner and direc-
information, tor of Wysowl Pty Ltd. in Queensland, Australia. He may be reached at john@wysowl.com.au. Brian K.
go to Nunnally, Ph.D., is associate director, Validation, at Wyeth Research. He may be reached at nunnalb@
gxpandjvt.com/bios wyeth.com. Bernard McGarvey, Ph.D., is process modeling group leader, Process Engineering Cen-
ter, Eli Lilly and Company. He may be reached at mcgarvey_bernard@lilly.com.

SFQSFTFOUTUIFSFBDUJPOQBUIXBZTJOBTZTUFNBOE
Figure 1: Powder fill control model structure.
DPOUBJOTUIFSFBDUJPOSBUFDPOTUBOUTUIBUNVTUCF Auger Feed System
FTUJNBUFECZEPJOHSFBDUJPOFYQFSJNFOUT.VDI
TUVEZIBTCFFOEPOFPOIPXUPHFUUIFCFTUQBSBN Control
FUFSFTUJNBUFTGSPNBHJWFOTFUPGFYQFSJNFOUBMEBUB Auger
Powder
Feed
Loop Speed

t'JSTUQSJODJQMFTNPEFMT5IFTFBSFNPEFMTXIFSF
QIZTJDBMSFMBUJPOTIJQTBSFCVJMUJOUPUIFNPEFM
BOEBMMUIFNPEFMQBSBNFUFSTSFRVJSFEUPVTFUIF Weighing System Controller
Vial Weight
NPEFMBSFBWBJMBCMFPSDBOCFGPVOEGSPNUFTUT
POTZTUFNTUIBUBSFVOSFMBUFEUPUIFTZTUFNCFJOH
TUVEJFE'PSFYBNQMF BGMVJENFDIBOJDTNPEFM Classifying S.P.C. Monitoring
System
PGGMPXJOBOBHJUBUFECBUDISFBDUPSXJMMOFFEUIF System
GMVJEWJTDPTJUZ*GUIFGMVJEJTBTUBOEBSEMJRVJE
TVDIBTXBUFS UIFWJTDPTJUZJTBWBJMBCMFJOUIFMJU
FSBUVSF*GUIFGMVJEJTOPUTUBOEBSE TJNQMFWJTDPT Figure 2: Comparison of uncontrolled and controlled
JUZNFBTVSFNFOUTDBOCFNBEFUPHFUUIFWBMVFT vial weights.
*OGJSTUQSJODJQMFTNPEFMJOH DPNQBSJTPOPGUIF 650
NPEFMSFTVMUTUPUFTUTPOUIFSFBMTZTUFNJTVTFE 625
Powder Weight
UPDPOGJSNUIFNPEFM OPUUPFTUJNBUFUIFNPEFM
600
QBSBNFUFST*OEFFEJUJTPOFPGUIFSFBMTVDDFTTFTPG
UIFTDJFOUJGJDNFUIPEUIBUGJSTUQSJODJQMFTNPEFMT 575
DBOCFCVJMUUIBUNBLFQSFEJDUJPOTUIBUBHSFFXJUI 550
BDUVBMTZTUFNCFIBWJPS 0 200 400 600 800 1000 1200
Time
5IJTQBQFSDPODFSOTUIFVTFPGQSPDFTTNPEFMTDPO 650
TJTUJOHPGNFDIBOJTUJDBOEGJSTUQSJODJQMFTNPEFMT OPU
Powder Weight
625
FNQJSJDBMNPEFMT BOEBOZSFGFSFODFTUPNPEFMTSFGFS
600
UPUIFGPSNFSBOEOPUUIFMBUUFS*OUIFOFYUQBSUPGUIJT
UXPQBSUTFSJFT TFWFSBMNPSFFYBNQMFTPGVTJOHQSPDFTT 575
NPEFMJOHUPSFEVDFWBSJBCJMJUZXJMMCFEFTDSJCFE 550
0 200 400 600 800 1000 1200
Time
WHY THE INCREASED
USE OF PROCESS MODELING? #FJOHBCMFUPFYQMBJOTZTUFNCFIBWJPSVTJOHQSPDFTT
5IFVTBHFPGQSPDFTTNPEFMTIBTJODSFBTFEESBNBUJDBMMZ NPEFMJOHJTBQPXFSGVMJOEJDBUPSUIBUXFVOEFSTUBOEUIF
PWFSUIFZFBST5IFSFBSFNBOZSFBTPOTGPSUIJT TZTUFNCFJOHNBOBHFEBUBGVOEBNFOUBMMFWFM/PUPOMZ
"TCVTJOFTTFTUSZUPLFFQUIFFOHJOFPGJNQSPWFNFOU DBOXFQSFEJDUXIBUXJMMIBQQFO CVUXIZJUXJMMIBQQFO
NPWJOH CFUUFSEFDJTJPOTOFFEUPCFNBEFXJUISFTQFDU 5IFDPNQVUFSQPXFSBWBJMBCMFUPQFPQMFUPTPMWF
UPXIFSFUPGPDVTUIFJSQSFDJPVTSFTPVSDFT FH QFPQMF QSPCMFNTJTJODSFBTJOHBMMUIFUJNF5IJTJTUSVFGPSCPUI
NPOFZ BOEUJNF
"MMCVTJOFTTFTSFBMJ[FUIBUTVDIEFDJ IBSEXBSFBOETPGUXBSF8FDBOTPMWFQSPCMFNTUPEBZ
TJPOTBSFNBEFXJUIVODFSUBJOUZ CVUJUXPVMECFHSFBUJG SPVUJOFMZUIBUZFBSTBHPXFSFUIFUPQJDTPG1I%UIF
UIFBNPVOUPGVODFSUBJOUZDPVMECFSFEVDFEPSBUMFBTUJG TFT/PUPOMZIBTUIFTDPQFPGQSPCMFNTUIBUXFDBOTPMWF
UIFZLOFXIPXNVDIVODFSUBJOUZXBTJOWPMWFE*OGBDU JODSFBTFECVUTPIBTUIFFGGJDJFODZXJUIXIJDIXFDBO
EFDJTJPOTDJFODFSFDPHOJ[FTUIBUUIFSFJTBWBMVFUPSFEVD TPMWFUIFN4FUUJOHVQBOETPMWJOHNPEFMTJTNVDIGBTUFS
JOHVODFSUBJOUZXJUIUIFDPODFQUPGUIFiWBMVFPGQFSGFDU UPEBZUIBUJUIBTCFFOJOUIFQBTUBOEUIJTUSFOEXJMMDPO
JOGPSNBUJPOw
UIJTSFQSFTFOUTUIFWBMVFUPBEFDJTJPO UJOVFVOBCBUFEJOUIFGVUVSF
NBLFSJGUIFZLOFXFYBDUMZXIBUUIFPVUDPNFXPVMECF "OPUIFSJNQPSUBOUFOBCMFSJTUIBUFEVDBUJPOBMJOTUJ
JGBHJWFOPQUJPOXBTDIPTFOGPSBEFDJTJPO UVUJPOTBSFJOUFHSBUJOHUIFOFXDBQBCJMJUJFTJOUPUIFJS
4ZTUFNTBSFCFDPNJOHNPSFDPNQMFYBMMUIFUJNF"T VOEFSHSBEVBUFDVSSJDVMB5IJTNFBOTUIBUOFXDPMMFHF
&JOTUFJOPODFTBJE i:PVDBOOPUTPMWFUIFQSPCMFNTPG HSBEVBUFTOPUPOMZIBWFUIFTLJMMTUPEPQSPDFTTNPEFM
UPEBZVTJOHUIFTBNFNFUIPETUIBUDSFBUFEUIFNJOUIF JOHCVUUIFZIBWFBOFYQFDUBUJPOUIBUCVTJOFTTFTGVSOJTI
GJSTUQMBDFw5IFPOMZXBZUPEFBMXJUIUIJTDPNQMFYJUZJT UIFNXJUIUIFDBQBCJMJUJFTUPFYQMPJUUIFTFTLJMMT
UPBQQMZNPSFTPQIJTUJDBUFEBQQSPBDIFTUPNBOBHFJU
JHOPSJOHJUBOEXJTIJOHGPSBTJNQMFSUJNFEPFTOUXPSL RELATIONSHIP BETWEEN
/PUFWFSZQSPCMFNDBOCFTPMWFEXJUIBTQSFBETIFFU"TB PROCESS MODELING AND
NBOBHFSPODFTBJE iPVSFGGPSUTBSFCFJOHTUJGMFECZQFP VARIABILITY REDUCTION
QMFBQQMZJOHTJNQMJTUJDTPMVUJPOTUPDPNQMFYQSPCMFNTw *OPSEFSUPTFFIPXQSPDFTTNPEFMJOHDBOCFVTFEUP

Figure 3: HPW distribution system. occurred.

5IJTNFBOTUIBUUIFSFBSFGPVSEJGGFSFOUTJUVBUJPOT
XIFSFWBSJBCJMJUZSFEVDUJPODBOUBLFQMBDF
t$PNNPODBVTFWBSJBCJMJUZQSFWFOUJPO
t4QFDJBMDBVTFWBSJBCJMJUZQSFWFOUJPO
HPW t$PNNPODBVTFWBSJBCJMJUZDPSSFDUJPO
STORAGE t4QFDJBMDBVTFWBSJBCJMJUZDPSSFDUJPO
TANK TO
USAGE
5IFGPMMPXJOHHJWFTFYBNQMFTPGFBDIUZQFPGBDUJWJUZ
POINTS TIPXJOHIPXQSPDFTTNPEFMJOHXBTVTFEBTQBSUPGUIF
WBSJBCJMJUZSFEVDUJPOBDUJWJUJFT
CASE STUDYCOMMON CAUSE

VARIABILITY AND PREVENTION
CONTROL "OFXQSPDFTTXBTCFJOHEFWFMPQFEUIBUSFRVJSFEB
VALVE(S) QPXEFSUPCFGJMMFEJOUPWJBMT5IFQPXEFSXBTGFEGSPN
PUMP BIPQQFSUPBOBVHFSTZTUFNUIBUGJMMTUIFWJBMTXIFO
JUTQJOT5IFBNPVOUPGQPXEFSGJMMFEEFQFOETPOUIF
BVHFSTQFFEUIFOVNCFSPGBVHFSSPUBUJPOTEVSJOHUIF
Figure 4: Pressure profile through valve. GJMMJOHBOEHJWFOBGJYFEGJMMUJNF1SJPSFYQFSJFODFIBE
CFFOXJUIMJRVJEGJMMJOH XIJDIEJGGFSTGSPNQPXEFSGJMM
STATIC PRESSURE PROFILE JOHJOPOFJNQPSUBOUBTQFDUMJRVJETIBWFBDPOTUBOU
EFOTJUZ XIFSFBTQPXEFSTDBODIBOHFUIFJSCVMLEFOTJUZ
XJUIJOBHJWFOCBUDI5IJTNFBOTUIBUUIFSFMBUJPOTIJQ
CFUXFFOUIFBVHFSTQFFEBOEUIFBNPVOUPGQPXEFS
GJMMFEDBOBOEEPFTWBSZPWFSUJNF4PEVSJOHUIFEFWFM
STATIC LIQUID PQNFOUPGUIFGJMMJOHQSPDFTT FYDFTTJWF DPNNPODBVTF

PRESSURE VAPOR WBSJBUJPOXBTTFFOJOUIFXFJHIUTGJMMFEJOUPUIFWJBMT
PRESSURE *UXBTEFUFSNJOFEUIBUUIJTMFWFMPGWBSJBCJMJUZXPVME
FLUID FLOW THROUGH THE CONTROL VALVE CFVOBDDFQUBCMFUPUIFQBUJFOUBOEXPVMEOFFEUPCF
SFEVDFETJHOJGJDBOUMZCFGPSFUIFQSPDFTTDPVMECFQVU
JONBOVGBDUVSJOH"OBVUPNBUFEXFJHIJOHTZTUFNXBT
SFEVDFWBSJBCJMJUZ MPPLBUWBSJBCJMJUZSFEVDUJPOPOUXP JEFOUJGJFEUIBUDPVMENFBTVSFUIFXFJHIUPGFBDIWJBMBU
EJNFOTJPOT QSPEVDUJPOTQFFET"TUIFBVHFSGFFETZTUFNXBTDPN
t5ZQFPGWBSJBCJMJUZ5IFSFBSFUXPUZQFTPGWBSJBCJMJUZ QVUFSDPOUSPMMFEBOEBTJHOBMXBTBWBJMBCMFUIBUDPVME
DPNNPODBVTFBOETQFDJBMDBVTFWBSJBCJMJUZ BEKVTUUIFBVHFSTQFFE UIFSFXBTIPQFUIBUBEEJOHUIJT

$PNNPODBVTFWBSJBCJMJUZJTEVFUPBDPOTUBOU XFJHIJOHTZTUFNXPVMEBMMPXBDPOUSPMTZTUFNUPCF
TZTUFNPGDBVTFTBOETPSFQSFTFOUTBTUBCMFQSPDFTT EFTJHOFEUIBUXPVMEBEKVTUUIFBVHFSTTQFFEBVUPNBUJ
4QFDJBMDBVTFWBSJBUJPOJTEVFUPBDIBOHJOHTZTUFN DBMMZBOENBJOUBJOUIFUBSHFUGJMMXFJHIU(JWFOUIBUUIJT
PGDBVTFTBOETPSFQSFTFOUTBOVOTUBCMFQSPDFTT BQQMJDBUJPOPGUIFBVUPNBUFEWJBMXFJHIJOHTZTUFNXBT
)PXUIFUBTLTBTTPDJBUFEXJUIWBSJBCJMJUZSFEVDUJPO OFXUPUIFDPNQBOZBOEUPUIFWFOEPS UIFSFXFSFOP
BSFQFSGPSNFEEFQFOETPOXIFUIFSUIFGPDVTJTPO EBUBBWBJMBCMFUPJOEJDBUFJGUIJTTUSBUFHZXPVMEXPSLPS
DPNNPOPSTQFDJBMDBVTFWBSJBUJPO OPU4PBNPEFMXBTCVJMUUPTJNVMBUFUIFDMPTFEMPPQ
t5JNJOHXIFUIFSUPQSFWFOUWBSJBUJPOCFGPSFUIF DPOUSPMPGUIFQSPDFTT BTTIPXOJO'JHVSF

QSPDFTTJTSVOPSDPSSFDUBGUFSUIFWBSJBUJPOIBTCFFO 5IFWJBMTXFSFGFEJOUPUIFXFJHIJOHTZTUFN BOEUIF
TFFOJOUIFQSPDFTT*OUIFGPSNFSDBTF BQSPDFTT NFBTVSFEXFJHIUTXFSFVTFEGPSUXPQVSQPTFTGJSTUMZ
or product design or proposed product or process UPDMBTTJGZUIFWJBMTBTBDDFQUBCMFPSOPUCZDPNQBSJOH
DIBOHFTIPVMECFFWBMVBUFEUPVOEFSTUBOEXIBU UIFQSPEVDUXFJHIUUPUIFTQFDJGJDBUJPOMJNJUT5IJTQSP
WBSJBCJMJUZDBOCFFYQFDUFEBOEJGUIFWBSJBUJPODBO UFDUFEUIFQBUJFOU4FDPOEMZ UIFXFJHIUXBTVTFEBTBO
CFSFEVDFEXJUIPVUIBWJOHUPTFFUIFWBSJBUJPOJO JOQVUUPUIFDPOUSPMMFSUIBUDPNQBSFEUIFXFJHIUUPUIF
UIFSVOOJOHQSPDFTT5IBUJT XFXBOUUPQSFWFOUUIF UBSHFUBOE CBTFEPOUIFEJGGFSFODF NPEJGJFEUIFBVHFS
WBSJBUJPOGSPNPDDVSSJOHJOUIFGJSTUQMBDF*OUIF TQFFE5IJTLFQUUIFWJBMTXFJHIUTXJUIJOTQFDJGJDBUJPO
MBUUFSDBTF UIFWBSJBCJMJUZTIPXTVQBTUIFQSPDFTT TPUIBUGFXFSXPVMECFSFKFDUFE5IJTSFEVDFEUIFMPTTFT
PSQSPEVDUJTNPOJUPSFEBGUFSUIFQSPDFTTJTSVO GSPNUIFQSPDFTTTPUIBUUIJTBQQSPBDIXPVMECFOFGJU
OJOH0ODFWBSJBCJMJUZJTTFFO NBLFBEFDJTJPOUP CPUIUIFQBUJFOUBOEUIFNBOVGBDUVSFS5IFNPEFMXBT
SFEVDFUIFWBSJBCJMJUZBOEUBLFBQQSPQSJBUFBDUJPOT VTFEUPUFTUTFWFSBMUZQFTPGDPOUSPMMFSBMHPSJUIN FH
5IBUJT XFDPSSFDUUIFWBSJBCJMJUZBGUFSJUIBTBMSFBEZ 1*% ..4& WFOEPSTPXOBMHPSJUIN
BOEUIFSFTVMUT

TIPXFEUIBUBTVJUBCMFDPOUSPMMFSDPVMECFEFTJHOFE
Figure 5: Linear to rotating table vial transport system.

'JHVSFTIPXTUIFJNQBDUPGUIJTDPOUSPMPOBTFU VIAL MOVEMENT
PGEBUB5IFUPQDIBSUTIPXTBTFUPGWJBMXFJHIUTXJUI
VIAL MOVEMENT
no control and the lower chart shows what this would
IBWFMPPLFEMJLFXJUIDPOUSPM5IFSFEMJOFJTUIFUBSHFU
GUIDE

5IFJNQSPWFNFOUJTDMFBS8JUIPVUUIFDPOUSPM RAIL
MFSJOQMBDF UIFNFBOXFJHIUXBTBOEUIFTUBOEBSE
EFWJBUJPOJTCVUXJUIUIFDPOUSPMMFSUVSOFEPO UIF
NFBOJTBOEUIFTUBOEBSEEFWJBUJPOJT5BHVDIJ
XPVMECFIBQQZWBSJBCJMJUZSFEVDFEBOENFBODMPTFS
UPUBSHFU*OEFFE GPSUIPTFGBNJMJBSXJUIUIF5BHVDIJ
MPTTGVODUJPO UIFMPTTGVODUJPOGPSUIFDBTFXJUIDPO
USPMXBTIBMGUIBUXJUIPVUDPOUSPM BIVHFJNQSPWFNFOU
JORVBMJUZ
5IFDPOUSPMTUSBUFHZJTJOUFSFTUJOHJO
UIBUJGZPVTUBSUXJUITUBUJTUJDBMQSPDFTTDPOUSPM 41$

BTBCBTJT ZPVXPVMEUFOEUPIBWFBEFBECBOEXJUIJO
XIJDIOPDPOUSPMMFSBDUJPOJTSFRVJSFE5IFOXIFO Figure 6: Model output showing a vial falling over.
ZPVHPPVUTJEFUIJTEFBECBOE ZPVUBLFBMBSHFDPOUSPM
BDUJPO*OQSBDUJDF BOEGSPNBUIFPSFUJDBMQPJOUPG
WJFXBTXFMM
UIJTTUSBUFHZJTOPUPQUJNBM*OGBDU UIF
PQUJNBMDPOUSPMTUSBUFHZGPSBOBVUPNBUJDDPOUSPMMFSBT
EFTDSJCFEJTUPIBWFOPEFBECBOEBOETNBMMFS BOE
NPSFGSFRVFOU
DPOUSPMBDUJPOT
/PUFBMTPPO'JHVSFUIBUFWFOXJUIBOBVUPNBUJD
DPOUSPMMFSJOQMBDF UIFSFJTSPPNGPSUSBEJUJPOBM41$5IJT
JTCFDBVTFBOZBVUPNBUJDDPOUSPMMFS OPNBUUFSIPXXFMM
EFTJHOFE IBTTPNFSFHJPOPVUTJEFPGXIJDIUIFDPOUSPM
EFHSBEFT FH XIFSFUIFQPXEFSCVMLEFOTJUZHFUTWFSZ
TNBMMPSWFSZMBSHF
.POJUPSJOHUIFXFJHIUTPOBDPOUSPM
DIBSUXJMMEFNPOTUSBUFUIJT*UTBCJHNJTUBLFUPUIJOL
UIBUBVUPNBUJDDPOUSPMBMPOFJTBMMZPVOFFE5IFTZTUFN MJLFMZTPVSDFPGUIFOPJTFBOEUIFOUPJNQMFNFOUDIBOHFT
XBTJNQMFNFOUFEJONBOVGBDUVSJOHBOETUBSUFEVQXJUI UPFMJNJOBUFUIFDBWJUBUJPO
PVUQSPCMFNT HJWJOHXFJHIUWBSJBUJPOBOEMPXSFKFDUSBUFT
JOMJOFXJUINPEFMQSFEJDUJPOT Cavitation
8IBUJTDBWJUBUJPOBOEIPXEPFTJUBSJTFJOBDPOUSPM
CASE STUDYSPECIAL CAUSE WBMWF 5IFCBTJDBOTXFSUPUIJTMJFTJOUIF#FSOPVMMJQSJO
VARIABILITY AND PREVENTION DJQMF
8IFOBGMVJEBDDFMFSBUFT JUTTUBUJDQSFTTVSF
1SPQFSEJTUSJCVUJPOPGIJHIQVSJUZXBUFS )18
JTJNQPS drops. This is the principle that leads to an upward lift
UBOUJOQIBSNBDFVUJDBMPQFSBUJPOT)18JTGFEGSPNB on an airplane wing when the shape of the wing causes
DFOUSBMTUPSBHFUBOLUPBMMUIFWBSJPVTVTBHFQPJOUTJO the air on the top of the wing to flow faster than the air
UIFNBOVGBDUVSJOHQSPDFTTVTJOHBQVNQUPQSPWJEFUIF PWFSUIFCPUUPNPGUIFXJOH*OBDPOUSPMWBMWF UIFMJR
NPUJWFGPSDF BTTIPXOJO'JHVSF VJEGMPXJOHUISPVHIUIFDPOUSPMWBMWFJTGPSDFEUISPVHI
8IFOOFFEFE )18JTTVQQMJFEUPBVTFQPJOUBUB BDPOTUSJDUJPOBTUIFDPOUSPMWBMWFQPTJUJPODIBOHFTUP
TQFDJGJFEGMPXSBUFVTJOHBDPOUSPMWBMWFUPLFFQUIFGMPX SFHVMBUFUIFGMPXSBUF5IJTDBVTFTUIFMJRVJETQFFEUP
SBUFBUUIFSFRVJSFETFUQPJOU*OPOFPGUIFTFEJTUSJCVUJPO JODSFBTFBOE IFODF UIFGMVJETUBUJDQSFTTVSFESPQT*GUIF
TZTUFNT JUXBTOPUJDFEUIBUBOPJTFDPVMECFIFBSEGSPN TUBUJDQSFTTVSFESPQTCFMPXUIFWBQPVSQSFTTVSF CVCCMFT
UIFDPOUSPMWBMWF4VDIOPJTFJTBOJOEJDBUPSPGDBWJUBUJPO XJMMCFGPSNFE JOFGGFDUUIFMJRVJECPJMT
0ODFUIF
*GUIJTDBWJUBUJPOJTOPUBEESFTTFE JUDBODBVTFQSFNBUVSF MJRVJEQBTTFTUISPVHIUIFSFTUSJDUJPO UIFMJRVJETQFFE
GBJMVSFPGUIFDPOUSPMWBMWF DBVTJOHQBSUJDVMBUFTGSPNUIF ESPQTBHBJOBOETPUIFTUBUJDQSFTTVSFSJTFT5IFCVCCMFT
WBMWFUPCSFBLPGGJOUPUIF)18 XIJDIJTIJHIMZVOEFTJS OPXDPMMBQTFBOEUIJTHJWFTBSJTFUPUIFDBWJUBUJPOFGGFDU
BCMF5IJTXPVMEDFSUBJOMZCFDMBTTJGJFEBTBTQFDJBMDBVTF 5IFDPMMBQTJOHCVCCMFTDBVTFWJCSBUJPOTUIBUBSFTUSPOH
FWFOU3BUIFSUIBOMFUTVDIBOFWFOUPDDVS JUJTQSFGFS FOPVHIUPDSFBUFTJHOJGJDBOUXFBSPOUIFDPOUSPMWBMWF
BCMFUPBEESFTTUIFDBVTFPGUIFDBWJUBUJPOBOEFMJNJOBUF DPNQPOFOUT XIJDIDBOMFBEUPQBSUJDVMBUFTJOUIF)18
JUTPUIBUUIFDPOUSPMWBMWFEPFTOPUFWFOUVBMMZGBJM4P BOEFWFOUVBMMZGVODUJPOBMGBJMVSFPGUIFDPOUSPMWBMWF
UIFBQQSPBDIVTFEXBTUPGJSTUDPOGJSNDBWJUBUJPOBTUIF 'JHVSFTIPXTBUZQJDBMTUBUJDQSFTTVSFQSPGJMFGPSUIF

)18BTJUQBTTFTUISPVHIUIFDPOUSPMWBMWF5IFDJSDV UJPONVTUCFDBSSJFEPVUJOBOBTFQUJDFOWJSPONFOU*O
MBSEPUTIPXTUIFQPJOUPGNJOJNVNTUBUJDQSFTTVSF*O QSBDUJDF UIFNPWJOHDPOWFZPSEPFTOPUHPSJHIUVQUPUIF
'JHVSF JUJTCFMPXUIFWBQPVSQSFTTVSF TIPXOCZUIF SPUBUJOHUBCMF5IFGFFEUBCMFOFYUUPUIFSPUBUJOHUBCMF
IPSJ[POUBMMJOF
BOETPDBWJUBUJPOJTFYQFDUFE EPFTOPUNPWFBOETPUIFWJBMTDPNJOHGSPNUIFNPWJOH
5IFNJOJNVNTUBUJDQSFTTVSFUIBUPDDVSTJOUIFDPO DPOWFZPSQVTIUIFWJBMTPOUPUIFGJYFEUBCMFBOEUIFO
USPMWBMWFJTBDPNQMFYGVODUJPOPGUIFIZESBVMJDTPGUIF POUPUIFSPUBUJOHUBCMF.PWJOHGSPNBMJOFBSNPUJPOUP
)18EJTUSJCVUJPOTZTUFN5IFSFGPSF BGJSTUQSJODJQMFT BSPUBUJOHNPUJPOJOUSPEVDFTJOTUBCJMJUZBOEWJBMTNBZ
NPEFMPGUIFEJTUSJCVUJPOTZTUFNXBTCVJMUXJUIUXP GBMMPWFS8IFOUIJTIBQQFOTUIFPQFSBUPSNVTUJOUFSWFOF
QVSQPTFTJONJOE BOEEFBMXJUIUIFGBMMFOWJBM5IJTJOUSPEVDFTJOFGGJDJFODZ
B5PDPOGJSNUIBUUIFTZTUFNBTDVSSFOUMZDPOGJHVSFE JOUPUIFNBOVGBDUVSJOHQSPDFTTBOESFEVDFTUISPVHIQVU
XPVMECFFYQFDUFEUPIBWFNJOJNVNTUBUJDQSFT *UBMTPJODSFBTFTUIFSJTLUPTUFSJMJUZBTTVSBODFCFDBVTFXF
TVSFTCFMPXUIFWBQPVSQSFTTVSFBOEIFODFUIBU XPVMEMJLFUIFPQFSBUPSTUPLFFQBXBZGSPNUIFTUFSJM
DBWJUBUJPOXPVMECFFYQFDUFE J[FEWJBMT FWFOUBLJOHJOUPBDDPVOUUIBUUIFPQFSBUPST
C"TTVNJOHDBWJUBUJPOJTDPOGJSNFEJO B
UPUFTUWBSJ BSFQSPQFSMZHPXOFEGPSBTUFSJMFNBOVGBDUVSJOHPQFSB
PVTPQUJPOTUPFMJNJOBUFUIFDBWJUBUJPOFGGFDU UJPO
5IJTWJBMGBMMEPXOIBQQFOTEVFUPUIFJOIFSFOU
5IFNPEFMJODMVEFEUIF)18TUPSBHFUBOL QVNQ JOTUBCJMJUZDBVTFECZUIFDIBOHFGSPNMJOFBSUPSPUBUJPOBM
DPOUSPMWBMWFT QJQJOH BOEBOZPUIFSSFMFWBOUDPNQP NPUJPOBOETPJTBDPNNPODBVTFWBSJBCJMJUZQSPCMFN
OFOUTTVDIBTQJQFFMCPXT BOEGMBOHFT5IFNPEFMEJE *OPSEFSUPVOEFSTUBOEUIFJNQPSUBOUQBSBNFUFSTUIBU
JOEFFEDPOGJSNUIBUUIFTZTUFNTIPVMEFYIJCJUDBWJUB DPOUSPMUIFTUBCJMJUZPGUIFWJBMT BNPEFMXBTCVJMUUIBU
tion. Then the following options were tested in the TJNVMBUFTUIFNPUJPOPGUIFWJBMTBTSJHJECPEJFT5IF
NPEFMUPTFFJGUIFZXPVMEJODSFBTFUIFNJOJNVNTUBUJD NBJOGFBUVSFTPGUIFTZTUFNUIBUXFSFMPPLFEBUDPOTJTUFE
QSFTTVSFBCPWFUIFWBQPVSQSFTTVSF PGUIFUPMFSBODFTPGUIFGJYFEBOESPUBUJOHUBCMFTWFSUJDBM
$IBOHFUIFUZQFPGDPOUSPMWBMWF EJTUBODFCFUXFFOUIFGJYFEBOESPUBUJOHUBCMFT EJGGFSFODF
1SFTTVSJ[FUIFTUPSBHFUBOL JOIFJHIUTPGCPUITJEFTPGUIFSPUBUJOHUBCMFTBOETPPO
" EEBOPSJGJDFBGUFSUIFDPOUSPMWBMWF*O 5IFNPEFMXBTBCMFUPTJNVMBUFTJUVBUJPOTXIFSFUIFWJBMT
effect his creates a restriction in the line after XPVMEUFOEUPCFNPSFPSMFTTTUBCMF0OFPGUIFOJDF
UIFDPOUSPMWBMWF XIJDIGPSDFTUIFQVNQUP GFBUVSFTPGUIFNPEFMJOHTPGUXBSFVTFEXBTJUTBCJMJUZUP
PQFSBUFBUBIJHIFSQSFTTVSFUPBDIJFWFUIF EJTQMBZUIFNPEFMSFTVMUTBTWJEFP'PSFYBNQMF 'JHVSF
TBNFGMPXSBUF TIPXTGPVSDPOTFDVUJWFGSBNFTPGBWJEFP SFBEJOHJO
%FDSFBTFUIF)18TVQQMZUFNQFSBUVSF UIFEJSFDUJPOPGUIFBSSPXT GPSBTJUVBUJPOXIFSFUIFWJBM
0QUJPOTPOFUISPVHIUISFFJODSFBTFUIFNJOJNVN GBMMTPWFS5IFSFTVMUPGUIJTNPEFMJOHXBTBOJNQSPWFE
TUBUJDQSFTTVSFJOUIFDPOUSPMWBMWFXIJMFPQUJPOGPVS VOEFSTUBOEJOHPGUIFSFRVJSFEUPMFSBODFT#FDBVTFUIF
SFEVDFTUIF)18WBQPVSQSFTTVSF8IJMFBMMUIFPQUJPOT OFXUPMFSBODFTXFSFJOUSPEVDFE UIFOVNCFSPGPQFSBUPS
DPVMEIBWFCFFODPOGJHVSFEUPFMJNJOBUFUIFDBWJUBUJPO JOUFSWFOUJPOTXFOUEPXOCZ"OEJUJTBOUJDJQBUFE
GPSQSBDUJDBMSFBTPOTPQUJPOUISFFXBTDIPTFOGPSJNQMF UIBUUIFTFUPMFSBODFTDBOCFUJHIUFOFEFWFOGVSUIFS
NFOUBUJPO5IFDBQBCJMJUZPGUIFNPEFMUPBMMPXPGGMJOF XIJDIXJMMMFBEUPGVSUIFSSFEVDUJPOTJOUIFOVNCFSPG
UFTUJOHJTBOPUIFSCPPOGPSUIFNPEFMJOHBQQSPBDI8JUI PQFSBUPSJOUFSWFOUJPOT
PVUUIFNPEFM UIFPQUJPOTXPVMEIBWFIBEUPCFUFTUFE
POUIFQSPEVDUJPOTZTUFN SFRVJSJOHTJHOJGJDBOUEPXO CASE STUDYSPECIAL CAUSE
UJNFBOEMFBEJOHUPVOIBQQZQSPEVDUJPONBOBHFNFOU VARIABILITY AND CORRECTION
5IFNPEFMXBTBCMFUPEFGJOFUIFSFRVJSFEPSJGJDFTJ[F 4UFSJMFGSFF[FESZJOH JF MZPQIJMJ[BUJPO
JTBDPNNPO
BOEXIFOJNQMFNFOUFE UIFOPJTFGSPNUIFDPOUSPMWBMWF VOJUPQFSBUJPOJOUIFQIBSNBDFVUJDBMJOEVTUSZ
*O
XBTIFBSEOPNPSF BOEOPOFPGUIFDPOUSPMWBMWFTGBJMFE UIJTPQFSBUJPO WJBMTGJMMFEXJUIMJRVJEBOEXJUITUPQQFST
KVTUSFTUJOHPOUPQBSFESJFECZGJSTUGSFF[JOHUIFNBUFSJBM
CASE STUDYCOMMON CAUSE BOEUIFOSFNPWJOHUIFMJRVJECZTVCMJNJOHUIFGSP[FO
VARIABILITY AND CORRECTION MJRVJEBUWFSZMPXWBDVVN5IFWJBMTTUBOEJOUSBZTPO
7JBMGJMMJOHJTBDPNNPOPQFSBUJPOJOQIBSNBDFVUJ NFUBMTIFMWFTJOUIFGSFF[FESZFS5IFOVNCFSPGUSBZTQFS
DBMNBOVGBDUVSJOH#FGPSFUIFWJBMTBSFGJMMFE UIFZBSF TIFMGWBSJFTGSPNQSPEVDUUPQSPEVDU0ODFUIFMJRVJEIBT
XBTIFEBOETUFSJMJ[FEJOBIFBUJOHUVOOFM5IFWJBMTFYJU CFFOSFNPWFE UIFTIFMWFTBSFGPSDFEEPXOCZBSBNBOE
UIFIFBUJOHUVOOFMWJBBDPOWFZPSUIBUNPWFTUIFWJBMTBT UIJTDBVTFTUIFTUPQQFSTUPCFQVTIFEJOUPUIFWJBMTTPUIBU
BHSPVQ)PXFWFS JOPSEFSUPGJMMUIFWJBMT UIFWJBMTNVTU BTFBMPDDVST5IJTLFFQTUIFNBUFSJBMJOUIFWJBMTTUFSJMF
CFPSHBOJ[FEJOUPBTUSBJHIUMJOF0OFXBZUIJTDBOCF BOETPJUJTDSVDJBMUIBUUIFTUPQQFSTEPOPUHFUEBNBHFE
BDDPNQMJTIFEJTUPNPWFUIFWJBMTPOUPBSPUBUJOHUBCMF EVSJOHUIJTTUPQQFSJOHQSPDFTT7JTVBMJOTQFDUJPOPGUIF
BOEVTFUIFDFOUSJGVHBMGPSDFBOEBHVJEFSBJMUPNPWFUIF TUPQQFSTJTDBSSJFEPVUCZPQFSBUPSTUPDPOGJSNUIBUUIF
WJBMTJOUPBTJOHMFGJMF BTTIPXOJO'JHVSF TUPQQFSJOHQSPDFTTJTBDDFQUBCMF
#FDBVTFUIFWJBMTIBWFBMMCFFOTUFSJMJ[FE UIJTPQFSB "GUFSBQBSUJDVMBSCBUDIXBTGSFF[FESJFE UIJTJOTQFD

UJPOTIPXFEBTNBMMOVNCFSPGEFGFDUJWFTUPQQFSTJO
Figure 7: Force per stopper histogram for batch with defects.
WJBMTUIBUXFSFMPDBUFEPOUIFMPXFTUTIFMG5IFQBSUJDVMBS
Simulation Results
EFGFDUJOEJDBUFEUIBUUIFWJBMTJORVFTUJPOTBXBMBSHFS 0.15 30000
UIBOOPSNBMTUPQQFSJOHGPSDF5IVT UIJTXBTBTQFDJBM
0.13 25000
DBVTFFWFOUGPSXIJDIBSPPUDBVTFBOEDPVOUFSNFBTVSFT
XFSFSFRVJSFE 0.10 20000
Relative Probability
5IFGJSTUUIJOHUIBUXBTOPUFEEVSJOHUIFJOWFTUJHB
Frequency
0.08 15000
UJPOXBTUIBUUIFOVNCFSPGUSBZTPOUIFCPUUPNTIFMG
XBTPOMZGJWFXIFSFOPSNBMMZOJOFUSBZTBSFQSFTFOU 0.05 10000
5IFSFBTPOGPSUIJTJTUIBUUIFCBUDITJ[FIBECFFOBMJUUMF
0.03 5000
MPXFS BOECFDBVTFTIFMWFTBSFMPBEFEUPQTIFMGGJSTU UIF
MBTUTIFMGHPUUIFMPXOVNCFSPGUSBZT#FDBVTFUIFTUPQ 0.00
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00
0
10.00 30.00 50.00 70.00 90.00 110.00 130.00 150.00
QFSJOHGPSDFJTEJTUSJCVUFEBDSPTTBMMUIFWJBMTPOUIFTIFMG
UIFTUPQQFSTPOUIFMPXFSTIFMGTBXBIJHIFSUIBOOPSNBM
GPSDF5IJTNFBOTUIBUUIFMPXFSOVNCFSPGUSBZTJTB
QMBVTJCMFFYQMBOBUJPOGPSUIFEFGFDUTCVUNPSFOFFEFEUP Figure 8: Force per stopper with normal number of trays.
CFEPOFUPGVMMZWFSJGZUIBUJUXBTUIFTPMFFYQMBOBUJPOGPS
UIFEFGFDUT*OQBSUJDVMBS JUEPFTOPUFYQMBJOXIZPOMZ Simulation Results
0.15 30000
BTNBMMOVNCFSPGWJBMTIBEEFGFDUJWFTUPQQFST*OPSEFS
UPGVMMZFYQMBJOUIFEFGFDUT JUJTOFDFTTBSZUPVOEFSTUBOE 0.13 25000
UIFWBSJBCJMJUZPGUIFTUPQQFSJOHQSPDFTT*UXBTLOPXO
Relative Probability
0.10 20000
UIBUBGPSDFJOUIFSFHJPOPGUPMCG QPVOEGPSDF
Frequency
0.08 15000
QFSTUPQQFSXPVMEDBVTFUIFEFGFDUTTFFOJOUIFCBUDI"
NPEFMXBTCVJMUUPTJNVMBUFUIFJNQBDUPGUIFWBSJBCJMJUZ 0.05 10000
JOBMMUIFTZTUFNBUUSJCVUFTUIBUBGGFDUUIFTUPQQFSJOHGPSDF
0.03 5000
TFFOCZBWJBM5IFTFJODMVEFEUIFGPMMPXJOH
t4UPQQFSJOHGPSDFBQQMJFECZUIFGSFF[FESZFSSBN 0.00 0
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00
t7JBM QSPEVDU TUPQQFSJOHSBN BOETIFMGXFJHIUT 10.00 30.00 50.00 70.00 90.00 110.00 130.00 150.00
t1SPEVDU WJBMTQFSUSBZ
t5SBZTQFSTIFMG
t#PXJOHPGUIFTIFMG Figure 9: Sensitivity analysis for the force per stopper
t0WFSBMMCPUUMFIFJHIUWBSJBCJMJUZ results shown in Figure 7.
t4UPQQFSGMBOHFUIJDLOFTTWBSJBCJMJUZ
".POUF$BSMPTJNVMBUJPOBQQSPBDIXBTVTFEUPQSF
-1.0 -0.8 -0.6 -0.4 -0.2 -0.0 0.2 0.4 0.6 0.8 1.0
EJDUUIFWBSJBCJMJUZJOUIFTUPQQFSGPSDFTFFOCZUIFWJBMT
Vial Height Delta 0.925
on the lowest shelf and the results presented as a histo
Stopper Flange Height 0.283
HSBN'JHVSFTIPXTUIFSFTVMUBOUIJTUPHSBNGPSUIFDPO Strain/Stress Curve Residual 0.138
EJUJPOTTFFOCZUIFWJBMTGPSUIFCBUDIXJUIEFGFDUT Bowing in Shelf 0.123
*UDBOCFTFFOUIBUUIFSFJTBTJHOJGJDBOUGSBDUJPOPGUIF Stress/Strain Curve Residual 0.075
IJTUPHSBNJOUIFSFHJPOPGMCGPSHSFBUFS5IFNPEFM % Drop in Stress at Edge of Shelf -0.058
XBTUIFOSVOXJUIUIFOPSNBMOVNCFSPGOJOFUSBZT5IF Shelf Weight 0.005
SFTVMUBOUIJTUPHSBNJTTIPXOJO'JHVSF/PXUIFQSPC Stoppering Pressure -0.002
BCJMJUZPGBWJBMTFFJOHBGPSDFBCPWFMCGJTTNBMM5IF Piston Diameter -0.002
Stopper Force Limit 0.001

GJOBMQJFDFPGFWJEFODFUIBUWBMJEBUFTUIFTNBMMOVNCFSPG
Stopper Flange Diameter 0.001
USBZTBTCFJOHUIFJTTVFSFTVMUTGSPNBTFOTJUJWJUZBOBMZTJT
Pressure Plate Weight 0.000
QFSGPSNFEPOUIFSFTVMUTTIPXOJO'JHVSF5IJTBOBMZTJT
JTTIPXOJO'JHVSF
This plot shows the correlation of each input to the
GPSDFQFSTUPQQFS XJUICMVFCBSTJOEJDBUJOHBQPTJUJWFDPS UPTUPQQFSTCFJOHNPSFDPNQSFTTFE
SFMBUJPOBOESFEJOEJDBUJOHBOFHBUJWFDPSSFMBUJPO5IFTF 5IFSFGPSF BTUVEZXBTDPOEVDUFEXIFSFUIFWJBMTPO
DPSSFMBUJPOWBMVFTDBOCFUSBOTGPSNFEUPHJWFUIFDPO UIFMPXFTUTIFMGXFSFDMFBOFEBOEUIFJSIFJHIUTNFB
USJCVUJPOPGFBDIJOQVUUPUIFUPUBMWBSJBCJMJUZPGUIFGPSDF TVSFE5IFO OFXTUPQQFSTXFSFQMBDFEPOUIFNBOEUIF
QFSTUPQQFS5IFDBMDVMBUJPOTIPXFEUIBUPGUIFWBSJ DPOEJUJPOTPGUIFEFGFDUJWFCBUDIXFSFSFQSPEVDFE5IF
BODFJOUIFGPSDF QFSTUPQQFS
EBUBXBTEVFUPWBSJBCJMJUZ WJBMTXFSFUIFOEJWJEFEJOUPUXPQPQVMBUJPOT POFXJUI
JOUIFWJBMIFJHIU XJUIWJBMTPGMBSHFSIFJHIUTCFJOHNPSF stopper defects and one with no defects. The popula
MJLFMZUPTIPXEFGFDUT7JBMTUIBUIBWFMBSHFSIFJHIUMFBE UJPOXJUIOPEFGFDUTIBEWJBMIFJHIUTJOUIFSBOHF

NNUPNN XIFSFBTUIPTFXJUIEFGFDUTIBEWJBM .D(BSWFZ # %BSHBU[ 3 %[JSCJL % i%JTDSFUF&WFOU4JNVMBUJPO

IFJHIUTNNUPNN BQQSPYJNBUFMZ/PUFUIBU 0GGFST3FBM7BMVF wChemical Processing Magazine +BOVBSZ
BMMWJBMTXFSFJOUIFTQFDJGJDBUJPOSBOHFPGNN 'PXMLFT 8JMMJBN: $SFWFMJOH $MZEF. i&OHJOFFSJOH.FUI
BMUIPVHIUIFWJBMTXJUIEFGFDUTXFSFEFGJOJUFMZPOUIF PETGPS3PCVTU1SPEVDU%FTJHOo6TJOH5BHVDIJ.FUIPETJO
IJHITJEFPGUIFTQFDJGJDBUJPOSBOHF5IJTTUVEZWBMJEBUFE 5FDIOPMPHZBOE1SPEVDU%FWFMPQNFOUw "EEJTPO8FTMFZ
UIFDBVTFPGUIFEFGFDUT5IFDPVOUFSNFBTVSFXBTUP .ZFST 3BZNPOE) Classical and Modern Regression with Ap-
NPEJGZUIFGSFF[FESZJOHQSPDFEVSFTPUIBUJOUIFGVUVSF plications OE&EJUJPO %VYCVSZ1SFTT
TIFMWFTXPVMEOPUCFMPBEFEXJUITPGFXUSBZT BOEUIF .POUHPNFSZCPPLPO%0&/&&%*/'0
QSPCMFNEJEOPUSFPDDVS 8BMUFS & 1SPO[BUP - Identification of Parametric Models from
Experimental Data 4QSJOHFS
CONCLUSION 8JOTUPO 8- 0QFSBUJPOT3FTFBSDI"QQMJDBUJPOTBOE"M
5IJTBSUJDMFTIPXTIPXQSPDFTTNPEFMJOHDBOCFVTFE HPSJUINT SE&EJUJPO 8BETXPSUI1VCMJTIJOH #FMNPOU
JOBMMQIBTFTPGWBSJBCJMJUZSFEVDUJPO1SPDFTTNPEFMJOH $BMJGPSOJB
DBOCFVTFEUPJEFOUJGZBSPPUDBVTF FMJNJOBUFQPUFOUJBM %FNJOH 8& Out of the Crisis QVCMJTIFECZ5IF$FOUFSGPS
SPPUDBVTFT JEFOUJGZQPUFOUJBMTPMVUJPOTBTXFMMBTUFTUUIF "EWBODFE&OHJOFFSJOH4UVEZ .*5 $BNCSJEHF .BTT
JNQBDUPGUIFTFTPMVUJPOTPGGMJOFXIFSFOPQSPEVDUJTQVU .D$POOFMM +PIO4 Analysis and Control of Variation UI&EJ
BUSJTL5IFVTFPGQSPDFTTNPEFMJOHBTBQPUFOUBMMZJOUIF UJPO QVCMJTIFECZUIF%FMBXBSF#PPLT
RVFTUGPSWBSJBCJMJUZSFEVDUJPOJTOPXXFMMFTUBCMJTIFE 8IFFMFS %POBME+ Understanding Variation, the Key to Managing
BOEJUTVTFXJMMDPOUJOVFUPHSPXBTNPEFMJOHUFDIOJRVFT Chaos OEFEJUJPO 41$1SFTT
BOEDPNQVUFSIBSEXBSFDPOUJOVFUPJNQSPWF %FM$BTUJMMP &OSJRVF Statistical Process Adjustment for Quality
Control 8JMFZ4FSJFTJO1SPCBCJMJUZBOE4UBUJTUJDT
REFERENCES .D$POOFMM +PIO4 /VOOBMMZ#SJBO BOE.D(BSWFZ #FSOBSE
.D(BSWFZ # i$SFBUF#VTJOFTT7BMVF5ISPVHI1SPDFTT.PEFM i.FFUJOH4QFDJGJDBUJPOT*T/PU(PPE&OPVHI5IF5BHVDIJ
JOH wChemical Engineering Progress W O QQ -PTT'VODUJPO wJournal of Validation Technology 4QSJOH
$IFO $ .D(BSWFZ # i.PEFMJOHBOE4JNVMBUJPOGPS1IBS )PMMBOE '" #SBHH 3 Fluid Flow for Chemical Engineers 4FD
NBDFVUJDBM1SPEVDUBOE1SPDFTT%FWFMPQNFOU w$PVODJMGPS POE&EJUJPO QVCMJTIFECZ&EXBSE"SOPME$PNQBOZ
$IFNJDBM3FTFBSDI1IBSNB8PSLTIPQ 8BTIJOHUPO %$ +VOF $BNFSPO 1 Good Pharmaceutical Freeze-Drying Practice *OUFS
QIBSN1SFTT *OD JVT

Paul L. Pluta
White Spots on Tablets

Compliance
IMAGEZOO/GETTY IMAGES Case Study #8
Paul L. Pluta
i$PNQMJBODF$BTF4UVEJFTwEJTDVTTFTDPNQMJBODF t.BOVGBDUVSJOHQFSTPOOFMJOEJDBUFEUIBUUIFSF
TJUVBUJPOTVTFGVMUPQSBDUJUJPOFSTJODPNQMJBODFBOE XBTQPUFOUJBMGPSJODPSSFDUMZBTTFNCMFENJMMJOH
WBMJEBUJPO&BDIDBTFQSFTFOUFEEFBMTXJUIBTQFDJGJD FRVJQNFOUDBVTJOHNBUFSJBMTUPOPUCFNJMMFE
DPNQMJBODFQSPCMFN FMFNFOUTPGXIJDIBSFEFTDSJCFEUP t$PNQBSBUJWFUFTUJOHPGDPOUSPMUBCMFUTBOE
EFNPOTUSBUFTUSBUFHZUPTPMWFDPNQMJBODFQSPCMFNT8F FYQFSJNFOUBMUBCMFUTVOEFSBDDFMFSBUFE
JOUFOEUIJTDPMVNOUPCFBVTFGVMSFTPVSDFGPSEBJMZXPSL UFNQFSBUVSFDPOEJUJPOTJOEJDBUFEUIBUUIF
BQQMJDBUJPOT5IFNBJOPCKFDUJWFPGUIJTDPMVNO6TFGVM VONJMMFEFYDJQJFOUXBTUIFDBVTFPGUIFEJTUJODU
BOEQSBDUJDBMJOGPSNBUJPO XIJUFTQPUTJOUIFUBCMFUT'%$#MVFXBT
3FBEFSDPNNFOUT RVFTUJPOT BOETVHHFTUJPOTBSF DIFNJDBMMZJODPNQBUJCMFXJUIUIFBMLBMJOF
OFFEFEUPIFMQVTGVMGJMMPVSPCKFDUJWFGPSUIJTDPMVNO excipient.
$BTFTUVEJFTJMMVTUSBUJOHDPNQMJBODFJTTVFTTVCNJUUFECZ t"OFXNJMMJOHFRVJQNFOUBTTFNCMZQSPDFEVSF
SFBEFSTBSFNPTUXFMDPNF1MFBTFTFOEZPVSDPNNFOUT BOEBOFXNJMMJOHQSPDFTTXBTEFWFMPQFE
BOETVHHFTUJPOTUPKPVSOBMDPPSEJOBUJOHFEJUPS4VTBO t5SBJOJOHPGNBOVGBDUVSJOHQFSTPOOFMXBT
)BJHOFZBUTIBJHOFZ!BEWBOTUBSDPN conducted.
t1SPDFTTWBMJEBUJPOXBTDPNQMFUFE
KEY POINTS DISCUSSED t"MMJOWFTUJHBUJPOT BOBMZTFT BOEDPODMVTJPOT
The following key points are discussed: XFSFEPDVNFOUFEUPDMPTFUIFDPSSFDUJWFBDUJPO
t"DBTFTUVEZJOWPMWJOHVOFYQFDUFEXIJUFTQPUTPO BOEQSFWFOUJWFBDUJPO $"1"
"MMJOWFTUJHBUJPO
BQSFWJPVTMZQSPCMFNGSFFCMVFUBCMFUQSPEVDUJT EPDVNFOUBUJPOXBTSFWJFXFECZ64'PPEBOE
EFTDSJCFE %SVH"ENJOJTUSBUJPOJOWFTUJHBUPST
t*OJUJBMTQFDVMBUJPOXBTUIBUUIFXIJUFTQPU t"EEJUJPOBMNPOJUPSJOHPGNBOVGBDUVSJOHXBT
QSPCMFNXBTFJUIFSHSFBTFTQPUTPSNJDSPCJBM DPOEVDUFEUPDPOGJSNTVDDFTTGVMDPSSFDUJWFBDUJPO
growth. t5IJTDBTFTUVEZGVSUIFSEFNPOTUSBUFEIJHIMZ
t'%$#MVFXBTUIFDPMPSBOUJOUIFUBCMFU VOMJLFMZDBVTFTPGBTFSJPVTDPNQMJBODFQSPCMFN
GPSNVMBUJPO BOJOBDUJWFFYDJQJFOUBOEJODPSSFDUFRVJQNFOU
t$IFNJDBMDIBOHF DIFNJDBMSFEVDUJPOJOBMLBMJOF setup.
Q)
PGUIFCMVFEZFXBTEFUFSNJOFEUPCFUIF t$PNQMJBODFQFSTPOOFMNVTULFFQBOPQFONJOE
source of the white spots. XJUIPVUQSFDPODFJWFECFMJFGTXIFOJOWFTUJHBUJOH
QSPCMFNTJUVBUJPOT
[
ABOUT THE AUTHORS
For more Author Paul L. Pluta, Ph.D., is a pharmaceutical scientist with extensive technical and management experience
information,
in the pharmaceutical industry. He is editor in chief of the Journal of GXP Compliance and the Journal of
Validation Technology. Dr. Pluta is also adjunct associate professor at the University of Illinois College of
go to
Pharmacy in Chicago, Illinois USA. He is the editor of and contributed chapters to Cleaning and Cleaning
gxpandjvt.com/bios Validation, published by PDA and Davis Healthcare International in 2009. He may be reached by e-mail at
paul.pluta@comcast.net.

Paul L. Pluta
INTRODUCTION WPMVNF3YQSPEVDU1SPEVDUBUUSJCVUFTBOEQSPDFTT
5IJTDBTFTUVEZXBTQSPWJEFEUPUIFJournal of GXP QFSGPSNBODFXFSFSFMJBCMFBOESFQSPEVDJCMF"OOVBM
ComplianceCZBSFBEFSXIPSFRVFTUFEBOPOZNJUZ5IF QSPEVDUSFWJFXTBOEQSPDFTTNPOJUPSJOHGPSTFWFSBM
FWFOUEFTDSJCFEJTBOBDUVBMPDDVSSFODF ZFBSTJOEJDBUFEOPPOHPJOHQSPCMFNTPSQSPDFTT
"TNBMMNPMFDVMFQIBSNBDFVUJDBMDPNQBOZIBE USFOET"MBSHFOVNCFSPGQSPEVDUDPNQMBJOUTXFSF
QSPEVDFEBOVODPBUFECMVFUBCMFUQSPEVDUGPSNBOZ VOFYQFDUFEMZSFDFJWFEGSPNDVTUPNFST$PNQMBJOUT
ZFBST"MBSHFOVNCFSPGQSPEVDUDPNQMBJOUTXFSF EFTDSJCFEFBTJMZWJTJCMFEJTUJODUXIJUFTQPUTPOUIF
VOFYQFDUFEMZSFDFJWFEGSPNDVTUPNFST$PNQMBJOUT UBCMFUQSPEVDU$PNQMBJOUTXFSFSFDFJWFEPOTFWFSBM
EFTDSJCFEEJTUJODUXIJUFTQPUTPOUIFCMVFUBCMFUQSPEVDU EJGGFSFOUCBUDIFTPGQSPEVDU CVUOPUBMMMPUTIBE
5IFDBVTFPGUIFQSPCMFNXBTOPUFBTJMZEFUFSNJOFE DPNQMBJOUT.BOVGBDUVSJOHQFSTPOOFMTQFDVMBUFE
4QFDVMBUJPOSBOHFEGSPNHSFBTFTQPUTUPNJDSPCJBM UIFDBVTFUPCFHSFBTFTQPUTGSPNUIFDPNQSFTTJOH
growth. QSPDFTT BQSPCMFNPDDBTJPOBMMZPCTFSWFEJOUIF
5IJTEJTDVTTJPOQSPWJEFT QBTU4BNQMFTPGQSPCMFNUBCMFUTXFSFSFDFJWFE
tProcess description background. The 4QFDVMBUJPOGSPNUIFBQQFBSBODFPGUIFTQPUTXBT
NBOVGBDUVSJOHQSPDFTTGPSUIFUBCMFUQSPEVDUJT UIBUNJDSPCJBMHSPXUIXBTPDDVSSJOHPOUBCMFUT/P
CSJFGMZEFTDSJCFE JOTUBODFTPGNJDSPCJBMHSPXUIIBEQSFWJPVTMZCFFO
tCompliance event"EFTDSJQUJPOPGUIFFWFOU UIF reported for this or any other products at the site.
LFZJTTVFTUPCFBEESFTTFE BOEBQQMJDBCMFDVSSFOU
HPPENBOVGBDUVSJOHQSBDUJDF $(.1
SFRVJSFNFOUT What are the Issues?
tInvestigation*OUFSWJFXTBOEBDUJPOTDPOEVDUFEUP 5IFSFXFSFTFWFSBMDSJUJDBMJTTVFTUPCFJOWFTUJHBUFE
JOWFTUJHBUFUIFFWFOU as follows:
tDiscussion,FZJOGPSNBUJPO BDUJWJUJFT t8IBUXBTUIFDPNQPTJUJPOPGUIFXIJUFTQPUT
and analysis. t8IBUXBTUIFTPVSDFPGUIFXIJUFTQPUT
tCorrective action and preventive action (CAPA). t8IZXFSFXIJUFTQPUTOPUEFUFDUFEJO
"DUJPOTBOEJNQSPWFNFOUTJNQMFNFOUFE UBCMFUJOTQFDUJPO
tPost CAPA-FTTPOTMFBSOFE.BJOUBJOJOHWBMJEBUJPO t8IZEJEUIFXIJUFTQPUDPNQMBJOUT
BOEQFSGPSNBODF TVEEFOMZPDDVS
t%JENBOVGBDUVSJOHPQFSBUPSTGPMMPXCBUDISFDPSE
PROCESS DESCRIPTION BACKGROUND EJSFDUJPOTQSPQFSMZ
5IFNBOVGBDUVSJOHQSPDFTTJOUIJTDBTFTUVEZJOWPMWFE
NBOVGBDUVSFPGBOVODPBUFECMVFUBCMFUQSPEVDU5IF CGMP Requirements
NBOVGBDUVSJOHQSPDFTTGPSUIFQSPEVDUDPNQSJTFE 3FMFWBOU$(.1
SFRVJSFNFOUTQPUFOUJBMMZ
BUZQJDBMXFUHSBOVMBUJPOQSPDFTT5IFGPMMPXJOH BQQMJDBCMFUPUIFBCPWFFWFOUBSFMJTUFEBTGPMMPXT
EFTDSJCFTUIFNBOVGBDUVSJOHQSPDFTT t4VCQBSU%o&RVJQNFOU
t5IFBDUJWFQIBSNBDFVUJDBMJOHSFEJFOU "1*
BOE t&RVJQNFOU%FTJHO 4J[F
PUIFSJOBDUJWFFYDJQJFOUTXFSFNJMMFEVTJOHBO BOE-PDBUJPO
JNQBDUNJMM t&RVJQNFOU$POTUSVDUJPO
t8FUHSBOVMBUJPOXBTQFSGPSNFEVTJOHBO t4VCQBSU'o1SPEVDUJPOBOE1SPDFTT$POUSPMT
BRVFPVTCJOEFSTPMVUJPODPOUBJOJOHB t4BNQMJOHBOE5FTUJOHPG*OQSPDFTT
QPMZNFSJDCJOEFS .BUFSJBMTBOE%SVH1SPEVDUT
t8FUHSBOVMBUJPOXBTESJFEVTJOHBGMVJEJ[FE t$POUSPMPG
CFEESZFS .JDSPCJPMPHJDBM$POUBNJOBUJPO
t%SJFEHSBOVMFTXFSFTJ[FEVTJOHBTJGUJOH t%SVH1SPEVDU*OTQFDUJPO
BQQBSBUVT0WFSTJ[FEHSBOVMFTXFSFNJMMFEVTJOH t4VCQBSU+o3FDPSETBOE3FQPSUT
BOJNQBDUNJMM t(FOFSBM3FRVJSFNFOUT
t3FNBJOJOHGPSNVMBUJPOJOHSFEJFOUTXFSFNJMMFE t#BUDI1SPEVDUJPOBOE
VTJOHBOJNQBDUNJMM5IJTJODMVEFE'%$#MVF $POUSPM3FDPSET
BTDPMPSBOUJOUIFGPSNVMBUJPO t1SPEVDUJPO3FDPSE3FWJFX
t"MMJOHSFEJFOUTXFSFNJYFEJOBUXJO
TIFMMCMFOEFS INVESTIGATION
t5BCMFUTXFSFDPNQSFTTFE *OWFTUJHBUJPOBOEVMUJNBUFSFTPMVUJPOPGUIJTFWFOU
SFRVJSFEJOWPMWFNFOUPGTFWFSBMHSPVQT5IFTFJODMVEFE
COMPLIANCE EVENT QFSTPOOFMJOWPMWFEJOUIFJODJEFOU NBOVGBDUVSJOH
"TNBMMNPMFDVMFQIBSNBDFVUJDBMDPNQBOZIBE BOERVBMJUZBTTVSBODF<2">
BOEUFDIOJDBMQFSTPOOFM
QSPEVDFEBOVODPBUFECMVFUBCMFUQSPEVDUGPS SFTQPOTJCMFGPSUIFNBOVGBDUVSJOHGPSNVMBUJPOBOE
NBOZZFBST5IFQSPEVDUXBTBSFMBUJWFMZMBSHF QSPDFTT5IFSFXFSFNBOZEFUBJMTUIBUOFFEFEUPCF

Paul L. Pluta
JOWFTUJHBUFEPSDPOGJSNFE1FSTPOOFMGSPNBMMHSPVQT XIJUFTQPUTXBTJEFOUJDBMUPUIFUBCMFUGPSNVMBUJPO
XFSFJOUFSWJFXFEBOEJOUFSBDUFEUPBEESFTTUIFBCPWF DPNQPOFOUT(SFBTFTQPUTXFSFFMJNJOBUFEBTBDBVTF
issues. PGUIFQSPCMFN
Changes in formulation materials. The purchasing
Product Testing IJTUPSZPGBMMQSPEVDUJOHSFEJFOUTXBTSFWJFXFE5IFSF
1SPEVDUTXJUIXIJUFTQPUTXFSFTVCNJUUFEGPS XFSFOPDIBOHFTJONBUFSJBMT5IFSFXFSFOPDIBOHFT
BOBMZUJDBMUFTUJOHGPSBMMQSPEVDUBUUSJCVUFT"MMUFTU JOTVQQMJFSTPGGPSNVMBUJPONBUFSJBMT"MMNBUFSJBMT
SFTVMUTXFSFBDDFQUBCMF1SPEVDUTXFSFGVMMZQPUFOU BOETVQQMJFSTIBEBHPPERVBMJUZIJTUPSZ.BUFSJBM
SFHBSEJOHBDUJWFESVH%JTTPMVUJPOUFTUEBUBXFSF DIBOHFTXFSFOPUBDBVTFPGUIFQSPCMFN
BDDFQUBCMFBOETIPXFEOPTMPXJOHPGUIFQSPEVDU FD&C Blue #2 chemistry. Technical personnel
EJTTPMVUJPOQFSGPSNBODF%FTQJUFUIFXIJUFTQPUT UIFSF JOWFTUJHBUFEUIFDIFNJDBMQSPQFSUJFTPG'%$
EJEOPUTFFNUPCFBOZBEWFSTFFGGFDUTPOUIFPUIFS #MVF UIFCMVFDPMPSBOUVTFEJOUIFGPSNVMBUJPO
QSPEVDUBUUSJCVUFT $IFNJDBMDIBOHF DIFNJDBMSFEVDUJPOJOBMLBMJOFQ)

PGUIFCMVFEZFXBTEFUFSNJOFEUPCFUIFTPVSDFPGUIF
Manufacturing Personnel Interviews XIJUFTQPUT'%$#MVFJTBXJEFMZVTFETZOUIFUJD
.BOVGBDUVSJOHQFSTPOOFMBGGJSNFEUIBUBMMCBUDI CMVFEZFBOEJTVTFEJOUIFEZJOHPGCMVFKFBOT
SFDPSEEJSFDUJPOTXFSFQFSGPSNFEBTTQFDJGJFE '%$#MVFJTBMTPLOPXOBT*OEJHPUJO *OEJHP
.BOVGBDUVSJOHPQFSBUPSTIBENBOVGBDUVSFEUIJT EZF BOE&JO&VSPQF*OBMLBMJOFQ) UIJTEZFJT
QSPEVDUNBOZUJNFTJOUIFQBTU5IFNBOVGBDUVSJOH DIFNJDBMMZSFEVDFEUPJOEJHPXIJUFPSMFVDPJOEJDP
TUBGGDPNQSJTFEFYQFSJFODFEPQFSBUPSTBOETPNFOFXMZ XIJDIJTXIJUFJODPMPS5IFGPSNVMBUJPOPGUIFCMVF
USBJOFEPQFSBUPST"MMQSPDFTTFTXFSFDPOTJEFSFEUP UBCMFUQSPEVDUDPOUBJOFEFYDJQJFOUTUIBUNBJOUBJOFE
CFXFMMFTUBCMJTIFEBOEQSPCMFNGSFF5IFSFXFSFOP BOBMLBMJOFQ)JOUIFUBCMFU5IJTQ)XBTOFDFTTBSZ
USBJOJOHJTTVFTPSWBSJBUJPOJOQFSGPSNBODFUIPVHIUUP GPS"1*QSPEVDUTUBCJMJUZ5IFBMLBMJOFQ)PGUIFUBCMFU
CFBTTPDJBUFEXJUIUIFNBOVGBDUVSJOHQSPDFEVSF DPSFDBVTFESFEVDUJPOPGUIFCMVFEZFUPJOEJHPXIJUF
XIJDIJOUVSODBVTFEXIJUFTQPSUT)PXFWFS CFDBVTF
QA Personnel Interviews PGUIFTVEEFODIBOHFJOUIFBQQFBSBODFPGUIFUBCMFU
2"QFSTPOOFMSFWJFXFEJOTQFDUJPOSFDPSETGSPN XIFOUIFSFIBECFFOOPDIBOHFTJOUIFGPSNVMBUJPO
QSPCMFNMPUT5BCMFUJOTQFDUJPOPDDVSSFEQFSJPEJDBMMZ NBUFSJBMT PSNBOVGBDUVSJOHQSPDFTTXBTVOLOPXO
UISPVHIPVUUIFDPNQSFTTJOHQSPDFTT-BCPSBUPSZ Manufacturing equipment.8IFOQSPCMFNT
QFSTPOOFMDPOEVDUFEBGJOBMUBCMFUJOTQFDUJPOPGB XJUIUIFCMVFEZFXFSFJEFOUJGJFE NBOVGBDUVSJOH
DPNQPTJUFTBNQMFJOUIFBOBMZUJDBMMBCPSBUPSJFT NBOBHFNFOUGVSUIFSQSPCFE'%$IBOEMJOHBOE
EJGGFSFOUQFPQMFJOEJGGFSFOUBSFBTJOTQFDUFEUIFUBCMFUT QSPDFTTJOH%JTDVTTJPOTXJUINBOVGBDUVSJOHPQFSBUPST
"MMUBCMFUJOTQFDUJPOSFDPSETXFSFBDDFQUBCMF/P JOEJDBUFEUIBUUIFSFXBTQPUFOUJBMGPSJNQSPQFSMZ
JOTUBODFTPGXIJUFTQPUTPOUBCMFUTXFSFSFQPSUFE BTTFNCMFENJMMJOHFRVJQNFOU*UXBTQPTTJCMFUIBU
UIFNJMMJOHTDSFFOTXFSFOPUDPSSFDUMZJOTUBMMFE*GUIJT
Technical Personnel Investigation PDDVSSFE VONJMMFEQBSUJDMFTXPVMEFTDBQFUIFNJMMJOH
5FDIOJDBMQFSTPOOFMDPOEVDUFEQBSBMMFMJOWFTUJHBUJPOT DIBNCFSBOEXPVMEOPUCFNJMMFE$PSSFDUFRVJQNFOU
BEESFTTJOHQPTTJCMFDBVTFTPGUIFXIJUFTQPUQSPCMFN TFUVQXBTOFDFTTBSZGPSHPPENJMMJOH
5IFTFJODMVEFENJDSPCJPMPHJDBMUFTUJOH FMFNFOUBM &YQFSJNFOUBMUSBJMTXFSFEFTJHOFEUPUFTUUIJT
BOBMZTJT DIBOHFTJOGPSNVMBUJPONBUFSJBMT BOE IZQPUIFTJT BTGPMMPXT
'%$#MVFDIFNJTUSZ"TJOWFTUJHBUJPOTJOEJDBUFE tControl milling.5IFJNQBDUNJMMXBTBTTFNCMFE
QPUFOUJBMDBVTFTGPSUIFQSPCMFN MFBETXPVMECF QSPQFSMZ"MLBMJOFFYDJQJFOUXBTTVDDFTTGVMMZ
pursued. NJMMFE4JFWFBOBMZTJTXBTDPOEVDUFEUP
Microbiological testing.8IJUFTQPUTXFSF DIBSBDUFSJ[FUIFNJMMFENBUFSJBM
SFNPWFEGSPNUIFUBCMFUTVSGBDFBOETVCNJUUFEGPS tProblem milling.5IFJNQBDUNJMMXBTBTTFNCMFE
NJDSPCJPMPHJDBMUFTUJOH/PNJDSPCJBMHSPXUIXBT JODPSSFDUMZ"MLBMJOFFYDJQJFOUXBTNJMMFE
PCTFSWFE5IFNPJTUVSFDPOUFOUPGUIFUBCMFUTXPVME 6ONJMMFEQPXEFSXBTPCTFSWFEFTDBQJOHUIF
OPUTVQQPSUNJDSPCJBMHSPXUI.JDSPCJBMHSPXUIXBT NJMMJOHDIBNCFS4JFWFBOBMZTJTXBTDPOEVDUFEUP
FMJNJOBUFEBTBDBVTFPGUIFQSPCMFN DIBSBDUFSJ[FUIFNJMMFENBUFSJBM%BUBJOEJDBUFE
Elemental analysis.8IJUFTQPUTXFSFSFNPWFE UIBUBTNBMMQFSDFOUBHFPGQBSUJDMFTXFSFOPU
GSPNUIFUBCMFUTVSGBDFBOETVCNJUUFEGPSNJDSPTDPQJD NJMMFE SFTVMUJOHJOSFMBUJWFMZMBSHFBMLBMJOF
FMFNFOUBMBOBMZTJT4BNQMFTPGHSFBTFTVTFEJOUIF QBSUJDMFTJOUIFUBCMFUDPSF*UXBTIZQPUIFTJ[FE
UBCMFUDPNQSFTTJOHQSPDFTTXFSFBMTPTVCNJUUFEGPS UIBUUIFTFMBSHFQBSUJDMFTSFBDUFEXJUICMVFEZFUP
DPNQBSJTPO/PFWJEFODFPGDPOUBNJOBUJPOCZHSFBTF DBVTFXIJUFTQPUTJOUIFCMVFUBCMFUT
XBTPCTFSWFE5IFFMFNFOUBMDPNQPTJUJPOPGUIF

Paul L. Pluta
tControl tablets.1SPEVDUUBCMFUTXFSF t%PDVNFOUBUJPO

NBOVGBDUVSFEXJUIDPSSFDUMZNJMMFENBUFSJBM t$PNNVOJDBUJPOUPPUIFSTJUFT
tProblem tablets.1SPEVDUUBCMFUTXFSF t0UIFSTJNJMBSQSPEVDUTNBOVGBDUVSFEBUUIFTJUF
NBOVGBDUVSFEXJUIJNQSPQFSMZNJMMFENBUFSJBM
$PNQBSBUJWFUFTUJOHPGDPOUSPMUBCMFUTBOE New Milling Equipment Assembly Procedure
FYQFSJNFOUBMUBCMFUTVOEFSBDDFMFSBUFEUFNQFSBUVSF 5IFJNQPSUBODFPGFRVJQNFOUBTTFNCMZJO
DPOEJUJPOTXBTUIFODPOEVDUFE4BNQMFTPGSFTQFDUJWF UIFNJMMJOHQSPDFTTJOEJDBUFEUIBUHSFBUFS
UBCMFUTXFSFBMTPTUPSFEVOEFSSFGSJHFSBUJPO5BCMFUT DPOUSPMXBTSFRVJSFE"OFXBOENPSFEFUBJMFE
NBOVGBDUVSFEXJUIUIFJNQSPQFSMZNJMMFENBUFSJBM BTTFNCMZQSPDFEVSFXBTJNQMFNFOUFE"MTP UIF
TUPSFEBUIJHIUFNQFSBUVSFEFWFMPQFEEJTUJODUMZWJTJCMF NBOVGBDUVSJOHTVQFSWJTPSDIFDLFEBOEBQQSPWFE
XIJUFTQPUT*UXBTDPODMVEFEUIBUUIFMBSHFQBSUJDMF TJHOBUVSF
UIFBTTFNCMZPGUIFNJMMCZUIF
TJ[FFYDJQJFOUXBTUIFDBVTFPGUIFEJTUJODUXIJUFTQPUT NBOVGBDUVSJOHPQFSBUPS4JHOPGGCZUIFTVQFSWJTPS
JOUIFUBCMFUT$PNQBSJTPOPGUBCMFUTTUPSFEVOEFS XBTBEEFEUPUIFNBOVGBDUVSJOHCBUDISFDPSE
BDDFMFSBUFEDPOEJUJPOTUPUIFSFGSJHFSBUFEUBCMFUT
JOEJDBUFEUIBUUIFSFGSJHFSBUFEUBCMFUTXFSFTMJHIUMZ New Milling Process
EBSLFSJODPMPSUIBOUIFIFBUFEUBCMFUT5IJTSFTVMU "OFXNJMMJOHQSPDFEVSFBOEDPOUSPMUFTUXBTBMTP
JOEJDBUFEUIBUUIF'%$#MVFXBTDIFNJDBMMZ JNQMFNFOUFEGPSUIJTQSPEVDU5IFBMLBMJOFFYDJQJFOU
JODPNQBUJCMFXJUIUIFBMLBMJOFFYDJQJFOU)PXFWFS BT UIBUXBTEFUFSNJOFEUPCFUIFDBVTFPGUIFXIJUF
MPOHBTUIFBMLBMJOFFYDJQJFOUXBTGJOFMZNJMMFE DPMPS TQPUQSPCMFNXBTNJMMFEUXJDFUISPVHIUIFJNQBDU
DIBOHFXBTTMPX EJGGVTF BOEHFOFSBMMZVOOPUJDFBCMF NJMM"GUFSNJMMJOH UIFNJMMFENBUFSJBMXBTTBNQMFE
to the naked or untrained eye. and tested for passage through an appropriate
TDSFFO"OZVONJMMFEQBSUJDMFTXPVMEOPUQBTT
DISCUSSION UISPVHIUIFTDSFFO*GVONJMMFEQBSUJDMFTXFSF
*OGPSNBUJPOPCUBJOFEUISPVHIJOUFSWJFXTBOE PCTFSWFE UIFQSPDFTTXBTTUPQQFEBOEBTVQFSWJTPS
TVCTFRVFOUFYQFSJNFOUBMXPSLFOBCMFEHPPE XBTJNNFEJBUFMZOPUJGJFE
VOEFSTUBOEJOHPGUIFQSPCMFN EFUFSNJOBUJPOPGUIF
SPPUDBVTF BOEVMUJNBUFDPSSFDUJWFBOEQSFWFOUJWF Process Validation
action. Original speculation as to potential causes of 1SPEVDUNBOVGBDUVSFEVTJOHUIFOFXNJMMJOHQSPDFTT
UIFQSPCMFNXFSFUFTUFEBOEEJTQSPWFO*OWFTUJHBUJWF XBTWBMJEBUFE5IFBNPVOUPGBMLBMJOFFYDJQJFOU
BOEFYQFSJNFOUBMXPSLCZUIFUFDIOJDBMHSPVQ DPNQSJTFEBWFSZMPXQFSDFOUPGUIFGPSNVMBUJPO
DPODMVTJWFMZEFUFSNJOFEUIFSPPUDBVTFPGUIFXIJUF 5IFBNPVOUPGQPUFOUJBMMZTNBMMFSQBSUJDMFTJ[F
TQPUQSPCMFN'%$#MVFEZFJOUIFGPSNVMBUJPO PGUIJTJOHSFEJFOUEJEOPUJNQBDUUIFQBSUJDMFTJ[F
XBTDIFNJDBMMZSFEVDFEUPJOEJHPXIJUF BXIJUF EJTUSJCVUJPOJOUIFCMFOEFEHSBOVMBUJPOBOEEJEOPU
TVCTUBODF-BSHFQBSUJDMFTPGBMLBMJOFFYDJQJFOU XIJDI BGGFDUUIFGMPXQSPQFSUJFTPSDPNQSFTTJOHQSPDFTTGPS
IBEOPUCFFONJMMFE DBVTFEUIFMPDBMJ[FEDIFNJDBM the product.
SFBDUJPOSFTVMUJOHJOUIFGPSNBUJPOPGUIFEJTUJODU
XIJUFTQPUT5IFBMLBMJOFFYDJQJFOUJOUIFGPSNVMBUJPO Training
XBTOPUNJMMFECFDBVTFUIFNJMMJOHTDSFFOTXFSF "MMQFSTPOOFMXFSFUSBJOFEPOUIFOFXFRVJQNFOU
JODPSSFDUMZJOTUBMMFEJOUIFJNQBDUNJMM$PNQBSBUJWF BTTFNCMZQSPDFEVSFBOENJMMJOHQSPDFTTQSPDFEVSF
UFTUJOHDPODMVTJWFMZQSPWFEUIBUUIFJNQSPQFSMZNJMMFE 5IFJNQPSUBODFPGUIFNJMMJOHQSPDFTTJO
alkaline excipient was the cause of the white spot NBOVGBDUVSJOHUIJTQSPEVDUBOEUIFDPSSFMBUJPOPG
QSPCMFN$PNQBSJTPOPGIFBUFEUBCMFUTUPSFGSJHFSBUFE JNQSPQFSNJMMJOHUPUIFXIJUFTQPUQSPCMFNXFSF
UBCMFUTDPOGJSNFEUIFTVTDFQUJCJMJUZPGUIF'%$#MVF XFMMDPNNVOJDBUFE5SBJOJOHPOQSPQFSTFUVQPG
UPBMLBMJOFDIFNJDBMSFEVDUJPO"TMPOHBTUIF NJMMJOHFRVJQNFOUXBTBQQMJDBCMFUPBMMQSPEVDUT
BMLBMJOFFYDJQJFOUXBTGJOFMZNJMMFE MJHIUFOJOHPGUIF NBOVGBDUVSFEBUUIFTJUF&TTFOUJBMMZBMMQSPEVDUT
CMVFUBCMFUTXBTOPUOPUJDFBCMF5IFJNQPSUBODFPGUIF JODMVEFENJMMJOHBTQBSUPGUIFJSNBOVGBDUVSJOH
NJMMJOHQSPDFTT BOEFTQFDJBMMZPGDPSSFDUJOTUBMMBUJPOPG process.
TDSFFOTJOUIFJNQBDUNJMM XBTDMFBSMZEFNPOTUSBUFE
Documentation
CORRECTIVE AND PREVENTIVE ACTIONS "MMXPSLBTTPDJBUFEXJUIUIFPSJHJOBMWBMJEBUJPO
5IFGPMMPXJOH$"1"BOEBTTPDJBUFEBDUJWJUJFT XBTDPNQMFUFE"MMJOWFTUJHBUJPOT BOBMZTFT BOE
were conducted: DPODMVTJPOTXFSFEPDVNFOUFEUPDMPTFUIF$"1"
t/FXNJMMJOHFRVJQNFOUBTTFNCMZQSPDFEVSF %FWFMPQNFOUPGUIFOFXNBOVGBDUVSJOHQSPDFTT
t/FXNJMMJOHQSPDFTT JODMVEJOHUSJBMSVOTXBTEPDVNFOUFEBOEGJMFEJO
t1SPDFTTWBMJEBUJPO UIFWBMJEBUJPOMJCSBSZBTTVQQPSUJOHJOGPSNBUJPOGPS
tTraining UIFWBMJEBUJPOQSPDFTTRVBMJGJDBUJPO 12
SVOT64

Paul L. Pluta
'PPEBOE%SVH"ENJOJTUSBUJPOJOWFTUJHBUPSTWJTJUFE XBTDIFNJDBMMZJODPNQBUJCMFXJUIUIFBMLBMJOF
UIFNBOVGBDUVSJOHTJUFJOSFTQPOTFUPDVTUPNFS FYDJQJFOU)PXFWFS BTMPOHBTUIFBMLBMJOFFYDJQJFOU
DPNQMBJOUT"MMJOWFTUJHBUJPOEPDVNFOUBUJPO XBTGJOFMZNJMMFE DPMPSDIBOHFXBTTMPX EJGGVTF
XBTSFWJFXFECZ'%"JOWFTUJHBUPST/P'%" BOEHFOFSBMMZVOOPUJDFBCMF5IFJNQPSUBODF
PCTFSWBUJPOTXFSFJTTVFE PGUIFNJMMJOHQSPDFTT BOEFTQFDJBMMZPGDPSSFDU
JOTUBMMBUJPOPGTDSFFOTJOUIFJNQBDUNJMM XBTDMFBSMZ
Communication to Other Sites EFNPOTUSBUFE
5IFJODJEFOUEFTDSJCFEIFSFJOXBTDPNNVOJDBUFEUP $"1"BOEBTTPDJBUFEBDUJWJUJFTDPOEVDUFE
PUIFSHMPCBMTJUFTMJDFOTFEUPNBOVGBDUVSFUIJTQSPEVDU JODMVEFEBOFXNJMMJOHFRVJQNFOUBTTFNCMZ
QSPDFEVSFUPQSFWFOUJNQSPQFSFRVJQNFOUTFUVQ B
Other Similar Products Manufactured At The Site OFXNJMMJOHQSPDFTT QSPDFTTWBMJEBUJPOPGUIFOFX
5IFNJMMJOHQSPCMFNEFTDSJCFEXBTVOJRVFUPUIF QSPDFTT USBJOJOHPGNBOVGBDUVSJOHQFSTPOOFM BOE
specific product. Other products did not contain EPDVNFOUBUJPOPGBMMBDUJWJUJFT5IFEPDVNFOUBUJPO
UIFTBNFDPMPSBOUTPSGPSNVMBUJPODPOUBJOJOHUIF XBTBMNPTUJNNFEJBUFMZVTFGVMXIFO'%"BVEJUFE
alkaline excipient. UIFTJUFJOSFTQPOTFUPDVTUPNFSDPNQMBJOUTUP
UIF"HFODZ5IJTFWFOUBOEJUTDPSSFDUJWFBDUJPOT
POST CAPA MONITORING XFSFDPNNVOJDBUFEUPPUIFSDPSQPSBUFTJUFT
"EEJUJPOBMNPOJUPSJOHPGNBOVGBDUVSJOHXBT NBOVGBDUVSJOHUIFTBNFQSPEVDU/PPUIFSQSPEVDUT
DPOEVDUFEUPDPOGJSNTVDDFTTGVMDPSSFDUJWFBDUJPO NBOVGBDUVSFEBUUIFTJUFXFSFTVTDFQUJCMFUPUIJT
"GJOBMSFQPSUTVNNBSJ[JOHBMMNPOJUPSJOHEBUB QSPCMFN"EEJUJPOBMNPOJUPSJOHPGNBOVGBDUVSJOH
XBTQSFQBSFEBGUFSUISFFNPOUIT NBOZMPUT
PG XBTDPOEVDUFE/PJOTUBODFTPGJNQSPQFSNBDIJOF
NBOVGBDUVSJOH/PJOTUBODFTPGJNQSPQFSNBDIJOF TFUVQXFSFPCTFSWFE/PNJMMJOHEJGGJDVMUJFTXFSF
TFUVQXFSFPCTFSWFE/PNJMMJOHEJGGJDVMUJFTXFSF SFQPSUFE/PDPNQMBJOUTPOQSPEVDUMPUTUIBU
SFQPSUFE/PDVTUPNFSDPNQMBJOUTPOQSPEVDUMPUT XFSFNBOVGBDUVSFEBGUFSUIFQSPDFTTDIBOHFTXFSF
UIBUXFSFNBOVGBDUVSFEBGUFSUIFQSPDFTTDIBOHFTXFSF SFDFJWFE$PSSFDUJWFBDUJPOXBTTVDDFTTGVM
SFDFJWFE$PSSFDUJWFBDUJPOXBTTVDDFTTGVM 5IJTDBTFTUVEZEFNPOTUSBUFEIJHIMZVOMJLFMZDBVTFT
PGBTFSJPVTDPNQMJBODFQSPCMFN$IFNJDBMQSPQFSUJFT
CONCLUSIONS PGJOBDUJWFJOHSFEJFOUTBSFOPUBUZQJDBMDBVTFPG
"DBTFTUVEZEFTDSJCJOHBDPNQMJBODFFWFOUJOXIJDI QSPCMFNT5IFNJMMJOHQSPDFTTJTBMTPOPUDPOTJEFSFE
BMBSHFOVNCFSPGQSPEVDUDPNQMBJOUTJOWPMWJOH UPCFBQSPCMFNBUJDNBOVGBDUVSJOHQSPDFTT5IFTFUXP
EJTUJODUXIJUFTQPUTPOBUBCMFUQSPEVDUXBT VOMJLFMZDBVTFTDPNCJOFEUPDBVTFBWFSZTJHOJGJDBOU
EJTDVTTFE5IFDBVTFPGUIFQSPCMFNXBTOPUFBTJMZ QSPCMFN$PNQMJBODFQFSTPOOFMNVTUCFBXBSFUIBU
EFUFSNJOFE4QFDVMBUJPOPOUIFQSPCMFNTPVSDF BMMJOHSFEJFOUTJOUIFGPSNVMBUJPO OPUKVTUUIFBDUJWF
SBOHFEGSPNHSFBTFTQPUTUPNJDSPCJBMHSPXUI5IF ESVH NBZIBWFTJHOJGJDBOUOFHBUJWFFGGFDUTPOUIF
QSPEVDUIBECFFOTVDDFTTGVMMZNBOVGBDUVSFEGPS QSPEVDU$PNQMJBODFQFSTPOOFMNVTUOPUBTTVNF
NBOZZFBSTXJUIPVUBOZJODJEFODFPGUIJTUZQFPG TJNQMFNBOVGBDUVSJOHQSPDFTTFTUPCFQSPCMFNGSFF
DPNQMBJOU5FDIOJDBMQFSTPOOFMDPOEVDUFEQBSBMMFM 5IFJOBDUJWFJOHSFEJFOUTBSFJNQPSUBOUDPNQPOFOUT
JOWFTUJHBUJPOTBEESFTTJOHQPTTJCMFDBVTFTPGUIF PGQSPEVDUT5IFQFSGPSNBODFPGFWFOQSFTVNBCMZ
XIJUFTQPUQSPCMFN5IFTFJODMVEFENJDSPCJPMPHJDBM NVOEBOFQSPDFTTFTTVDIBTNJMMJOHNVTUCFTFSJPVTMZ
UFTUJOH FMFNFOUBMBOBMZTJT DIBOHFTJOGPSNVMBUJPO DPOTJEFSFE$PNQMJBODFQFSTPOOFMNVTULFFQBOPQFO
NBUFSJBMT BOEDIFNJTUSZPG'%$#MVF UIF NJOEXJUIPVUQSFDPODFJWFECFMJFGTXIFOJOWFTUJHBUJOH
DPMPSBOUJOUIFGPSNVMBUJPO$IFNJDBMDIBOHF QSPCMFNTJUVBUJPOT
DIFNJDBMSFEVDUJPOJOBMLBMJOFQ)
PGUIFCMVFEZF
XBTEFUFSNJOFEUPCFUIFTPVSDFPGUIFXIJUFTQPUT
"MMPUIFSQPUFOUJBMQSPCMFNDBVTFTXFSFEJTQSPWFO REFERENCE
.BOVGBDUVSJOHQFSTPOOFMEFUFSNJOFEUIBUUIFSF '%" $PEFPG'FEFSBM3FHVMBUJPOT 5JUMF 'PPEBOE%SVHT
XBTQPUFOUJBMGPSJNQSPQFSMZBTTFNCMFENJMMJOH 1BSU $VSSFOU(PPE.BOVGBDUVSJOH1SBDUJDFGPS'JOJTIFE
FRVJQNFOUDBVTJOHNBUFSJBMUIBUXBTOPUNJMMFE 1IBSNBDFVUJDBMT Federal Register 4FQU
5IJTJOUVSODBVTFEGPSNBUJPOPGUIFXIJUFTQPUT GXP
&YQFSJNFOUBMUSJBMTDPOGJSNFEUIJTIZQPUIFTJT
5IJTXPSLBMTPJOEJDBUFEUIBUUIF'%$#MVF
0SJHJOBMMZQVCMJTIFEJOUIF"VUVNOJTTVFPGThe Journal of GXP Compliance

D. Wylie McVay, Jr.
Compliance Case Study #5

Secondary Packages With
IMAGEZOO/GETTY IMAGES Defective Glue Joints
D. Wylie McVay, Jr.
KEY POINTS DISCUSSED PDDVSSFODF5FDIOPMPHJDBMBEWBODFTNBZBMUFSTPNF

The following key points are discussed in this article: EFUBJMT CVUOPUUIFQSJODJQMFTSFMBUFEUPUIFDPSSFDUJWF
t"DPNQMJBODFDBTFTUVEZJOWPMWJOHEFGFDUJWF BDUJPOTBOEQSFWFOUJWFBDUJPOT $"1"

packages is discussed 5IJTEJTDVTTJPOQSPWJEFTUIFGPMMPXJOH
t"NVUVBMUSVTUJOHSFMBUJPOTIJQXJUIUIFWFOEPSJT tProcess descriptions. Background on
LFZUPQSPCMFNSFTPMVUJPO NBOVGBDUVSJOHQSPDFTTFTJOWPMWFE
t&NQIBTJTJTQMBDFEPOUIFJNQPSUBODFPG tCompliance event."EFTDSJQUJPOPGUIFFWFOU UIF
VOEFSTUBOEJOHTVQQMJFSBOEDVTUPNFSDBQBCJMJUJFT LFZJTTVFTUPCFBEESFTTFE BOE
t5IFFGGFDUJWFOFTTPGRVBMJUZDPOUSPMTJTNPTUPGUFO BQQMJDBCMFDVSSFOUHPPENBOVGBDUVSJOH
the result of careful attention to details QSBDUJDF $(.1
SFRVJSFNFOUT
t2VBMJUZDPOUSPMPGBQSPDFTTNVTUCFQMBDFEBUUIF tInvestigation.*OUFSWJFXTBOEBDUJPOT
critical control step DPOEVDUFEUPJOWFTUJHBUFUIFFWFOU
t4BNQMFJOTQFDUJPOJTBQPPSEFUFSNJOBOUPGUIF tDiscussion.,FZJOGPSNBUJPO
MFWFMPGJODPNJOHRVBMJUZ BDUJWJUJFT BOEBOBMZTJT
t1SPDFTTEFGFDUTBOEEFGFDUJWFVOJUTBSFNPTU tCAPA."DUJPOTBOEJNQSPWFNFOUT
often generated during startup and shutdown of JNQMFNFOUFE5IFTFJODMVEFOFXPQFSBUJOH
FRVJQNFOU QSPDFEVSFT USBJOJOH OFXTBNQMJOH JNQBDUPO
t&GGFDUJWFQSPDFTTDPOUSPMTBSFEFWFMPQFEXJUIJOUIF QBTUNBOVGBDUVSJOH BOEBTTPDJBUFEBDUJWJUJFT
DPOUFYUPGNBOVGBDUVSJOHQSJODJQMFT tPost CAPA.7FSJGZJOHBOENBJOUBJOJOH
t-FTTPOTMFBSOFEGSPNUIJTDBTFTUVEZ XIJDIXFSF QBDLBHJOHMJOFQFSGPSNBODF
DSJUJDBMUPTPMWJOHUIFQSPCMFN BSFOPUUZQJDBMMZ tOther activities.*NQSPWFNFOUT
GPVOEJOBRVBMJUZNBOVBMPSUSBJOJOHMFDUVSF JNQMFNFOUFEBTSFTVMUPGLOPXMFEHFHBJOFE
BOEBDUJWJUJFTBTTPDJBUFEXJUIUIFBCPWF
INTRODUCTION JODMVEJOHUIFTJUFBQQSPBDIUPTBNQMJOH
5IJTDBTFTUVEZXBTQSPWJEFEUPUIFJournal of GXP
ComplianceCZBSFBEFSXIPSFRVFTUFEBOPOZNJUZ*U PROCESS DESCRIPTIONS
QSPWJEFTPUIFSTXJUIJOTJHIUTJOUPIPXUPSFDPHOJ[F 5IJTDPNQMJBODFFWFOUJOWPMWFEUXPTFQBSBUFQSPDFTTFT
BOESFTPMWFMPXMFWFMRVBMJUZQSPCMFNTUIBUOFFEMFTTMZ 5IFGJSTUXBTSVOBUUIFDBSUPONBOVGBDUVSFS5IF
JODSFBTFPQFSBUJOHDPTUWJBNBOVGBDUVSJOHEPXO TFDPOEXBTSVOBUUIFQIBSNBDFVUJDBMNBOVGBDUVSFS
UJNF5IFDJSDVNTUBODFTQSPWJEFEEFTDSJCFBOBDUVBM VTJOHDBSUPOTGSPNUIFDBSUPOTVQQMJFS&BDIQSPDFTT
[
ABOUT THE AUTHORS
For more Author D. Wylie McVay, Jr. is a consultant in pharmaceutical manufacturing, quality, and regulatory affairs.
information, He is a Certified Quality Engineer, Certified Quality Auditor, and is certified in regulatory affairs. He has
go to published or presented on numerous quality topics including effective implementation of lifecycle control
gxpandjvt.com/bios strategies, aseptic fill, terminal sterilization, dosage form quality including tablets, capsules, liquids, and
other topics. He may be reached by e-mail at dwmcvayjr@gmail.com.

D. Wylie McVay, Jr.
QSPWJEFEWJUBMDMVFTBCPVUEFGFDUGSFRVFODZBOE KVTUQFOFUSBUFUIFFYBDUUIJDLOFTTPGUIFDIJQCPBSE
TPVSDF5IF$"1"MFEUPCFUUFSQSPDFTTDPOUSPMTBUUIF TUPDL5IFDVUUJOHLOJGFJTBMTPiOJDLFEwUPDSFBUF
TVQQMJFSTQMBOUBOECFUUFSIBOEMJOHPGDBSUPOTCZUIF JOUFOUJPOBMNJOPSEBNBHFUPUIFDVUUJOHFEHF
QIBSNBDFVUJDBMNBOVGBDUVSFS5IFDPNQMJBODFFWFOU 5IFEBNBHFEBSFBXJMMOPUDPNQMFUFMZTFWFSUIF
DPNQSJTFEMPXGSFRVFODZJOUFSNJUUFOUDBSUPOEFGFDUT QSJOUFEDIJQCPBSETUPDL5IJTTNBMMBNPVOUPG
5IFQIBSNBDFVUJDBMNBOVGBDUVSFSGJSTUCFDBNFBXBSF DIJQCPBSEGJCFSLFFQTUIFDBSUPOTBUUBDIFEUP
PGUIFQSPCMFNGSPNMPXFGGJDJFODZSBUJOHTPOUIF the sheet until the cartons are separated prior to
production line. The carton supplier was unaware that HMVJOH4FQBSBUJOHUIFDBSUPOTSFTVMUTJOWFSUJDBM
UIFSFXBTBQSPCMFNEVFUPQPPSPSJODPSSFDUQSPCMFN TUBDLTPGUIFTBNFQSJOUQPTJUJPO
EFTDSJQUJPOT5IFJOTUJODUJWFSFTQPOTFGPSUIFTVQQMJFS Process control alert:4FUVQTDSBQNVTUCF
BOEUIFNBOVGBDUVSFSXBTUPiJOTQFDUPVUUIFEFGFDUJWF DPOUSPMMFECZUIFTVQQMJFSBOENBZJODMVEF
DBSUPOTw5IFCFUUFSTPMVUJPOXBTUPDBSFGVMMZFYBNJOF QSJOUJOHEFGFDUT(MVJOHCZPSEFSFEQSJOUQPTJUJPO
UIFQSPCMFNBOEiCVJMERVBMJUZJOwCZDPSSFDUJOHUIF FOIBODFTUIFDVTUPNFSTBCJMJUZUPUSPVCMFTIPPU
QSPDFTTPSQSPDFTTDPOUSPMT"HFOFSBMPWFSWJFXPG BOEDPOUSPMQSJOUJOHQSPCMFNTUIBUXFSFOPU
UIFTFQSPDFTTFTGPMMPXTUPQSPWJEFUIFSFBEFSXJUIB EJTDPWFSFECZUIFTVQQMJFS
general understanding of each process. tGluing.5IFHMVJOHFRVJQNFOUJTBQQSPYJNBUFMZ
GFFUJOMFOHUIGSPNUIFDBSUPOGFFETJEF
Carton Supplier Process UPUIFGJOBMHMVFEDBSUPOUBLFPGG4UBDLTPG
.BOVGBDUVSFPGTFDPOEBSZQBDLBHJOH GPMEJOHDBSUPOT
DBSUPOCMBOLT QSFGFSBCMZCZQSJOUJOHQPTJUJPO
DPNQSJTFEUIFGPMMPXJOHQSPDFTTTUFQT BSFMPBEFEJOUPBWFSUJDBMGFFENBHB[JOFPGUIF
tSheet.1SFQBSFDBSUPODIJQCPBSETUPDL5IF HMVJOHNBDIJOF5IFDBSUPOTBSFSFNPWFEPOFBU
DBSUPOTUPDLJTQSPWJEFECZUIFQBQFSNJMMJO BUJNFGSPNUIFCPUUPNPGUIFTUBDL TJNJMBSUP
KVNCPSPMMT XIJDIJTUIFODVUJOUPUIFQSPQFS UIFGFFEPGBQSJOUFS*OEJWJEVBMDBSUPOTBSFiIFME
GMBUTIFFUTJ[FGPSQSJOUJOH5IJTJTSFGFSSFEUPBT JOQMBDF wTBOEXJDIFECFUXFFODPOUJOVPVT MPPQ
iTIFFUJOHw GFFEUFOTJPOCFMUT"TUIFDBSUPOQSPDFFETEPXO
Process control alert:%JSFDUJPOBMPSJFOUBUJPO UIFCFMU TIBQFESBJMTHFOUMZCFOEUIFDBSUPOBUUIF
PGUIFTIFFUTNVTUCFNBJOUBJOFEGPSGJCFS TDPSFMJOFTBOEQPTJUJPOUIFHMVFKPJOUGBDFUPUIF
HSBJOPSJFOUBUJPOBOEDPOTJTUFOUTIFFUTJ[F HMVFBQQMJDBUPS*OPMEFSNBDIJOFT UIFHMVFXBT
tPrint.1SJOUUIFGMBUTIFFUT.BDIJOFTFUVQGPS BQQMJFECZBTPMJENFUBMXIFFMUIBUSFWPMWFEJOB
registering and aligning print colors generates TIBMMPXQPPMPGXIJUF XBUFSCBTFEHMVF(MVFJT
TDSBQUIBUDPOUBJOTEFGFDUJWFMBCFMJOHTVDIBT BQQMJFEUPUIFKPJOUBOEGPMEFEPWFSUPDPOUBDUUIF
NJTTJOHQSJOU5PBTTVSFDPOTJTUFOUQSJOUMPDBUJPO PQQPTJUFGBDFPGUIFDBSUPOUPDPNQMFUFUIFKPJOU
POBMMTIFFUT UIFTIFFUTBSFSFHJTUFSFEXJUIBMBUFSBM DPOOFDUJPO$BSUPOTBSFIFMEJOQMBDFBTUIFHMVF
force to a stationary side rail to assure the printing sets.
JTMPDBUFEJOUIFTBNFQPTJUJPOFWFSZUJNFSFMBUJWF Process control alert:.BDIJOFTFUVQDBSUPOT
UPUIJTFEHF'MBUTIFFUTBSFGFETFRVFOUJBMMZBOE TIPVMECFTDSBQQFE BTUIFSFXJMMCFEFGFDUT
DPOUJOVPVTMZUISPVHIUIFQSJOUJOHQSFTT&BDI QSFTFOU(MVFEFGFDUTNBZCFDSFBUFEEVSJOH
TIFFUJTQSJOUFEJOBiTUFQBOESFQFBUwGBTIJPO TUBSUVQPSTIVUEPXO&GGFDUJWFRVBMJUZDPOUSPM
BTTIPXOJOUIF'JHVSF*GUIFSFMBUJWFQPTJUJPO TIPVMECFDBSFGVMMZEFTJHOFEUPQSFWFOU
SFHJTUSBUJPOTIJGUTEVSJOHQSJOUJOH UIFQSJOUUPDVU JOUFSNJYJOHPGHPPEBOEEFGFDUJWFDBSUPOT
UPMFSBODFTDBOCFDPNQSPNJTFEBOEDPVMEMFBEUP tBulk package.(MVFEDBSUPOTBSFQBDLBHFEGPS
DVUPGGDPQZPSNJTTJOHDPQZ TIJQNFOUUPUIFQIBSNBDFVUJDBMNBOVGBDUVSFS
Process control alert:4FUVQTDSBQFDPOUBJOT
MBCFMJOH QSJOU
EFGFDUT1SJOUTDSBQNBZCFVTFE Pharmaceutical Manufacturing Process
GPSDVUUJOHTFUVQBOENVTUCFDPOUSPMMFE 5IFQSPDFTTBUUIFQIBSNBDFVUJDBMNBOVGBDUVSFS
tCut.1SJOUFETIFFUTBSFEJFDVUBOETDPSFECZ JOWPMWFEGPSNJOHBOEGJMMJOHPGUIFCMJTUFSQBDLBHF
BIJHIQSFDJTJPOTUBNQJOHQSPDFTTTJNJMBSUPB QSJNBSZQBDLBHJOHGPMMPXFECZJOTFSUJPOPGUIFGJMMFE
DPPLJFDVUUFS5IFDVUUJOHEJFCPBSEHFOFSBMMZ CMJTUFSTJOUPUIFTFDPOEBSZQBDLBHFDBSUPOTPCUBJOFE
DPOTJTUTPGBIJHIRVBMJUZQMZXPPEiIPMEFSwUIBU GSPNUIFDBSUPONBOVGBDUVSFSBTEFTDSJCFEQSFWJPVTMZ
JTMBSHFSUIBOUIFQSJOUFETIFFU8JUIJOUIFFEHFT 4QFDJGJDBMMZUIJTQSPDFTTDPNQSJTFEUIFGPMMPXJOHTUFQT
PGUIFCPBSE QSFDJTJPOMBTFSDVUTJOUIFTIBQFPG tBlister formation.5IFSNPGPSNJOHNBUFSJBMJO
UIFGJOBMGMBUDBSUPOIBWFCFFONBEF5IFTFDVUT UIJTDBTFJTQPMZWJOZMDIMPSJEF 17$
5IFSPMMJT
PSDIBOOFMT BSFCVSOFEUISPVHIUIFCPBSETPUIBU NPVOUFEPOBIPMEFSBUPOFFOEPGUIFQBDLBHJOH
SB[PSTIBSQDVUUJOHLOJWFTBOEVOTIBSQFOFETDPSF NBDIJOF5IF17$JTIFBUFEUPNBLFJUQMJBCMF
CMBEFTDBOCFNPVOUFEJOUPUIFDIBOOFM5IFEJF 5IFIFBUFE17$JTGFEUISPVHIGPSNJOHEJFTUIBU
JTNPVOUFEUPUIFDVUUJOHNBDIJOFBOEBEKVTUFEUP XJMMGPSNUIFJNQSFTTJPOTPSDBWJUJFTUIBUXJMMIPME

D. Wylie McVay, Jr.
Figure: UIFNBDIJOF5IFTFDPOEBSZQBDLBHFTXFSFUIFOGJMMFE
Typical printed sheet (dark area is waste XJUIUXPCMJTUFSTUSJQTPGQIBSNBDFVUJDBMUBCMFUT"T
material). UIFQSPDFTTQSPHSFTTFE UIFGJMMFEDBSUPOTXFSFGPVOE
UPIBWFFJUIFSOPHMVFPSFYDFTTJWFBNPVOUTPGESJFE
HMVF CPUIPGXIJDISFTVMUFEJONBDIJOFEBNBHFBOE
QSPEVDUJPOEPXOUJNF5IFQSPCMFNXBTJOUFSNJUUFOU
BOEBMPXMFWFMEFGFDUSBUFEFGFDUTPDDVSSFEJOQBSUT
QFS )PXFWFS FWFOUIJTMPXMFWFMEFGFDUSBUFPG
DBSUPORVBMJUZMFEUPDPTUMZEPXOUJNFBOETJHOJGJDBOU
FYDFTTJOWFOUPSZPGQBDLBHJOHNBUFSJBMTKVTUUPBTTVSF
UIBUBEFRVBUFGPMEJOHDBSUPOTXFSFBMXBZTBWBJMBCMF
GPSTDIFEVMFEQSPEVDUJPOSVOT"EEJUJPOBMMZ UIF
RVBMJUZDPOUSPMMBCPSBUPSZTQFOUTJHOJGJDBOUBNPVOUTPG
OPOWBMVFBEEJOHUJNFPOBEEJUJPOBMJOTQFDUJPO5IF
JODSFBTFEJOTQFDUJPOSBSFMZJEFOUJGJFEQSPCMFNTCFGPSF
UIFDBSUPOTXFSFOFFEFEGPSQSPEVDUJPO.BOVGBDUVSJOH
UIFESVHQSPEVDU5IF17$JTIFMECZUIFFEHFTUP PQFSBUJPOTQFSTPOOFMXFSFGSVTUSBUFECZNJTTJOHLFZ
transport the product through the processing. EFMJWFSZUBSHFUT*OTVNNBSZ UIFTJUVBUJPOQSFTFOUFE
tFill blisters with tablets.%SVHQSPEVDUJTQMBDFE BOPQQPSUVOJUZUPTJHOJGJDBOUMZJNQSPWFUIFPWFSBMM
JOUPFBDIDBWJUZBDDPSEJOHUPUIFQSPEVDUEFTJHO CVTJOFTTQFSGPSNBODF
5BCMFUTUIBUIBWFMPXPSOPGSJBCJMJUZQSPCMFNTNBZ
CFGMPPEGFEPWFSUIFPQFOCMJTUFSDBWJUJFT4PGUFS What Were The Issues?
UBCMFUTNBZCFJOTFSUFEJOEJWJEVBMMZUPQSFWFOU 5IFSFXFSFTFWFSBMDSJUJDBMJTTVFTUPCFJOWFTUJHBUFE BT
EBNBHF follows:
tSeal blisters.-JEEJOHTUPDLTVDIBTBMVNJOVN t8IZEJETPNFCPYFTIBWFOPHMVF 8IBU
GPJMJTVOSPMMFEPWFSUIF17$CMJTUFSDPOUBJOJOH XBTUIFGSFRVFODZPGUIJTPDDVSSFODF
UBCMFUTJOBDPOUJOVPVTGFFENBOOFS1SFTTVSFBOE t8IZEJETPNFCPYFTIBWFFYDFTTJWFBOEESJFE
DPOUBDUXJUIIFBUJOHXIFFMTPSQMBUFOTXJMMXBSN HMVF 8IBUXBTUIFGSFRVFODZPGUIJTPDDVSSFODF
UIFQBDLBHFBOEBDUJWBUFUIFTFBMCFUXFFOUIFCBDL t8FSFUIFUXPPDDVSSFODFTDPOOFDUFE
MJEEJOHBOEUIFGJMMFECMJTUFSDBWJUZ PSTFQBSBUFFWFOUT
tCut blisters.5IFTFBMFEQBDLBHFJTEJFDVUBOE t8IZXFSFUIFTFQBDLBHFTOPUEFUFDUFE
TFQBSBUFEGSPNUIF17$XFC JORVBMJUZVOJUTBNQMJOH
tInsert blister into carton.5IFDVUCMJTUFSTUSJQT t8IZXBTUIJTIBQQFOJOHPOBWBMJEBUFEQSPDFTT
BSFMPBEFEUPBGFFENBHB[JOF5IFGPMEJOHDBSUPOT VTJOHWBMJEBUFEBOERVBMJGJFEFRVJQNFOU
GSPNUIFTVQQMJFSBSFMPBEFEJOUPBTFQBSBUF t8BTUIFQSPCMFNBTVQQMJFSQSPCMFNPSB
NBHB[JOF5IFQBDLBHJOHNBDIJOFXJMMPQFO NBOVGBDUVSJOHQSPCMFN
the carton and seal one end. The carton is then
GJMMFEXJUIUXPCMJTUFSTUSJQT*OUIJTDBTF UIF CGMP Requirements
DBSUPOTXFSFGJMMFEXJUICMJTUFSTUSJQTXIJMFJO 3FMFWBOU(.1SFRVJSFNFOUTBQQMJDBCMFUPUIFBCPWF
UIFIPSJ[POUBMQPTJUJPO5IFDBSUPOTIBWFUPCF FWFOUBSFMJTUFEBTGPMMPXT
QFSGFDUMZHMVFECZUIFTVQQMJFSPUIFSXJTF UIF t$'3E4VCQBSU&$POUSPM
NBDIJOFXJMMKBNEVSJOHPQFOJOHBOEDSFBUF PS$PNQPOFOUTBOE%SVH1SPEVDU
EPXOUJNF5IJTQSPDFTTTUFQJTTJHOJGJDBOUMZ $POUBJOFSTBOE$MPTVSFT5FTUJOHBOE
BGGFDUFECZEFGFDUJWFDBSUPOT BQQSPWBMPSSFKFDUJPOPGDPNQPOFOUT
tShipping case packaging. The filled cartons are ESVHQSPEVDUDPOUBJOFST BOEDMPTVSFT
QBDLBHFEJOUPCVMLDBSUPOTGPSTIJQQJOH t$'3B4VCQBSU(1BDLBHJOH
BOE-BCFMJOH$POUSPM.BUFSJBMT
COMPLIANCE EVENT FYBNJOBUJPOBOEVTBHFDSJUFSJB
5IFDPNQMJBODFFWFOUPDDVSSFEBUUIFQIBSNBDFVUJDBM
NBOVGBDUVSFSVTJOHTFDPOEBSZDBSUPOTPCUBJOFEGSPN INVESTIGATION
UIFDBSUPOTVQQMJFS5IFFWFOUJOWPMWFEEFGFDUJWF *OWFTUJHBUJPOBOEVMUJNBUFSFTPMVUJPOPGUIJTFWFOU
TFDPOEBSZDPOUBJOFSTVTFEUPQBDLBHFQIBSNBDFVUJDBM SFRVJSFEJOWPMWFNFOUPGTFWFSBMHSPVQTUISPVHI
UBCMFUTPOBOJOUFHSBUFEIJHITQFFECMJTUFSGPSNJOH BTVQQMJFSDFSUJGJDBUJPOJOJUJBUJWF1BDLBHJOH
TFBMJOHBOEDBSUPOGJMMJOHQBDLBHJOHMJOF5IFPCTFSWFE FOHJOFFSJOH FOHJOFFSJOHTVQQPSU QVSDIBTJOH
EFGFDUXBTPWFSHMVFEPSVOHMVFEDBSUPOTCFJOHGFEJOUP NBOVGBDUVSJOHQFSTPOOFMBOERVBMJUZBTTVSBODF

D. Wylie McVay, Jr.
QFSTPOOFMUFBNFEUPBTTFTTUIFTVQQMJFSDPOUSPMTBT DIBOHFTUIBUDPVMEBGGFDUGVODUJPOJOHPGUIFBQQSPWFE
XFMMBTBTTFTTUIFUSVFTQFDJGJFEOFFETPGUIFUBCMFU EFTJHO5IFDPNQBOZXBTBCMFUPTIFEBQQSPYJNBUFMZ
packaging process. PGFYDFTTJOWFOUPSZCZFMJNJOBUJOHFYDFTT
6OEFSUIFVNCSFMMBPGUIFTVQQMJFSDFSUJGJDBUJPO safety stock.
QSPDFTT BNVUVBMMZBHSFFBCMFEBUFGPSBTJUFBVEJU
XBTEFUFSNJOFE5IFRVBMJUZBTTVSBODF 2"
DISCUSSION
NFNCFSTQSPWJEFEUIFMPHJTUJDBMTVQQPSUBOE *OGPSNBUJPOPCUBJOFEUISPVHIJOUFSWJFXTBOE
DPNNVOJDBUFEBTBTJOHMFQPJOUPGDPOUBDUUPUIF TVCTFRVFOUGPMMPXVQBVEJUTBOETJUFNFFUJOHTFOBCMFE
supplier. The intent of the audit was clearly stated: HPPEVOEFSTUBOEJOHPGUIFQSPCMFNBOEMFEUPUIF
to learn the folding carton process in order to GJOBMTPMVUJPO.PTURVBMJUZQSPCMFNTEPOPUSFRVJSF
VOEFSTUBOEUIFLFZGBDUPSTPGRVBMJUZDPOUSPM5IF BEWBODFENBUIFNBUJDBMUFDIOJRVFT CVUSBUIFSDBSFGVM
UFBNTIPQFXBTUPMFBSOBOTXFSTUPUIFQSFDFEJOH BOEJOTJHIUGVMPCTFSWBUJPO5IJTTJUVBUJPOXBTOP
RVFTUJPOTGSPNUIFTVQQMJFSTQFSTQFDUJWF*O FYDFQUJPO*UXBTPOMZUISPVHIBTFUPGiGSFTIFZFTwBOE
PUIFSXPSET EFWFMPQVOEFSTUBOEJOHCFUXFFOUXP BNVUVBMVOEFSTUBOEJOHPGFBDIPUIFSTOFFETUIBUUIF
CVTJOFTTQBSUOFST UFBNXBTBCMFUPEFWFMPQBDMFBSDPOUFYUPGUIFSPPU
'PMMPXJOHTUBOEBSEBVEJUJOHQSPUPDPM BEWBODF cause.
DPNNVOJDBUJPODMBSJGJFEUIFPCKFDUJWFTBOEMPHJTUJDTPG 4JNQMZQVU UIFSPPUDBVTFPGVOHMVFEPSPWFSHMVFE
UIFUSJQ5IFQVSQPTFXBTFTUBCMJTIFEBTJEFOUJGJDBUJPO DBSUPOTXBTQBSUJBMPSDPNQMFUFESZJOHPGHMVFXIFO
PGTPVSDFTPGMPXMFWFMBOEJOUFSNJUUFOUDBSUPOHMVF UIFNBDIJOFXBTTIVUEPXOGPSCSFBLTBOEMVODI
KPJOUGBJMVSFT5IFJNQBDUPGJNQSPQFSMZHMVFEDBSUPOT 5IFTPMVUJPOJOWPMWFEUXPBDUJPOTCZUIFDBSUPO
XBTEFTDSJCFEJOUFSNTPGEPXOUJNF NBDIJOF NBOVGBDUVSFS'JSTU UIFNBDIJOFPQFSBUPSTXPVME
EBNBHF BOEBTTPDJBUFEEJTSVQUJPOUPPOUJNFEFMJWFSZ DBSSZUIFHMVJOHDZDMFUPBMPHJDBMQSPDFTTIPMEJOHQPJOU
5IFTJUFBTTFTTNFOUEBZBSSJWFEBOEPVSUFBNNFU CFGPSFTUPQQJOHUIFNBDIJOFUPUBLFBCSFBL4FDPOE
XJUIUIFTVQQMJFSFBSMZJOUIFNPSOJOH*OJUJBMEJBMPHVF BGJYFEBNPVOUPGDBSUPOTXFSFSFNPWFEGSPNUIF
CFUXFFOUIFQBSUJFTDFOUFSFEPOCFUUFSVOEFSTUBOEJOH NBDIJOFVQPOTUBSUVQ5IFFOEPGUIFHMVFDZDMFXBT
UIFTVQQMJFSTQSPDFTTBOEDPOUSPMT FTUBCMJTIJOHMJOFT iGMBHHFEwTPUIBUTVTQFDUDBSUPOTGSPNUIFOFYUTUBSUVQ
PGEJSFDUDPNNVOJDBUJPOGPSDMFBSBOEUJNFMZGPMMPX XFSFSFNPWFEGSPNUIFQBDLBHJOHMJOF FGGFDUJWFRVBMJUZ
VQPOJTTVFT BOEEFWFMPQJOHBSFQPSUDBSEUPHBVHF control).
QSPHSFTT0GDPVSTF QBSUOFSTIJQTBSFBUXPXBZTUSFFU -PPLJOHCBDL UIFTUBSUVQUIBUGPMMPXFECSFBLTPS
5IFUFBNXBTHMBEUPIPTUUIFDBSUPOTVQQMJFSUPTIPX MVODIJONBOZSFTQFDUTXBTBNVTJOHUPPCTFSWF5IF
IPXUIFUFBNIBOEMFE PSNJTIBOEMFE
UIFDBSUPOT BQQMJDBUPSXIFFM XIJDISFWPMWFEJOBTIBMMPXSFTFSWPJS
JOTUPSBHFBOEPOUIFQSPEVDUJPOMJOF5IJTNVUVBM PGHMVF EFWFMPQFEBESJFETLJOPGHMVFEVSJOHUIF
VOEFSTUBOEJOHQSPWFEUPCFUIFLFZUPDBOEJEBOE TIPSUCSFBLUJNFXIFOSPUBUJPODFBTFE5IFSFTVMUXBT
FGGFDUJWFEJBMPHVFPGUIFJTTVFTSBUIFSUIBOiQPJOUJOHPG EFGFDUJWFDBSUPOT4PNFUJNFTPOTUBSUVQ UIFXIFFM
GJOHFSTPGCMBNFw TMVOHUIFCMPCPGHMVFJOUPUIFDBSUPOBOETFBMFEJU
4VDDFTTSFRVJSFEUJNF QBUJFODF BOEQFSTJTUFOU TIVU"UPUIFSUJNFT UIFHMVFPOUIFXIFFMIBEESJFE
QVSTVJUPGFYDFMMFODFPOUIFQBSUPGCPUIQBSUOFSTUP BOEUIFGJSTUDBSUPOPSUXPEJEOPUSFDFJWFXFUHMVF
EFWFMPQBOPQFOBOEBMJHOFEWJFXPGUIFQSPCMFN POUIFKPJOU0OMZDBSFGVMPCTFSWBUJPODPODMVTJWFMZ
5IFGJSTUTFWFSBMNPOUITDPVMECFUFSNFEBQFSJPEPG identified this critical control point.
iCVJMEJOHUSVTUwXIFSFPOMZTVTQJDJPOIBEQSFWJPVTMZ
FYJTUFE.BOZPGUIFBTTVNQUJPOTUIFDVTUPNFSIFME Sampling Procedure
UVSOFEPVUUPCFJODPSSFDU'PSFYBNQMF UIFQBDLBHJOH 0OFPGUIFLFZQPJOUTUPCFMFBSOFEJTBMNPTUBMM
engineers were certain precision die cutting was the QSPDFTTFTEPOPUHFOFSBUFEFGFDUTSBOEPNMZ"TUIJT
SPPUDBVTFPGUIFQSPCMFNT5IJTBTTVNFEDPODMVTJPO DBTFTIPXT EFGFDUTXFSFDSFBUFEBUDMFBSMZEFGJOFE
BDUVBMMZEFMBZFEJEFOUJGJDBUJPOPGUIFLFZRVBMJUZ UJNFTEVSJOHUIFNBOVGBDUVSJOHQSPDFTT TVDIBTTUBSU
BUUSJCVUF"OJNQPSUBOUMFTTPOXBTMFBSOFE5IFSFJT VQPSTIVUEPXO$POUSPMBOEDBQUVSFPGEFGFDUJWF
OPTVCTUJUVUFGPSDBSFGVMTUVEZPGBQSPCMFN VOJUTJTBDSJUJDBMFMFNFOUPGBTVDDFTTGVMRVBMJUZDPOUSPM
"GUFSOFBSMZBZFBSPGDBSFGVMSFWJFX UIFFGGPSUTQBJE TUSBUFHZ3PUFJOTQFDUJPORVBMJUZTZTUFNT XIJDI
PGG5IFTVQQMJFSEFWFMPQFECFUUFSQSPDFTTDPOUSPMT UIF USFBUBMMUIFGJOJTIFEDBSUPOTBTBIPNPHFOFPVTMPU
UFBNMFBSOFEIPXUPQSPQFSMZTUPSFDBSUPOTUPQSFWFOU MBDLUIFTFOTJUJWJUZUPQSPQFSMZJOWFTUJHBUFUIJTLJOE
EBNBHF :FT DBSUPOTEFGPSNXIFOUPQMPBEFEXJUI PGRVBMJUZQSPCMFN*UJTFBTZUPTFFUIBUBSBOEPN
GPVSGVMM IFBWZQBMMFUT%FGPSNFEDBSUPOTEPOPUXPSL TBNQMFBQQSPBDIBTTVSFTEFGFDUTXJMMCFEFMJWFSFEUP
QSPQFSMZ
#PUIQBSUJFTEJTDPWFSFEUIFJNQPSUBODF UIFDVTUPNFS%PFTJUOPUNBLFCFUUFSQSPDFTTDPOUSPM
PGBOFGGFDUJWF XFMMDSBGUFERVBMJUZBHSFFNFOUXJUI TFOTFUPSFNPWFLOPXOEFGFDUTXIFOUIFZPDDVS
SFTQPOTJCJMJUZGPSCPUIUIFDVTUPNFSBOETVQQMJFS 5IFDVTUPNFSTXJMMBQQSFDJBUFUIJT
UPLFFQUIFPUIFSJOGPSNFEPGNBKPSUFDIOPMPHZ

D. Wylie McVay, Jr.
CORRECTIVE ACTIONS AND PREVENTIVE JNQSPWFNFOUXBTCFUUFSEFGFDUDBQUVSFEVSJOH

ACTIONS NBDIJOFTUBSUVQBGUFSCSFBLTBOEBGUFSMVODI
$"1"BDUJWJUJFTXFSFTQFDJGJFEBUCPUIUIFDBSUPO *OUFSFTUJOHMZ TUBSUVQTDSBQXBTXFMMDPOUSPMMFEEVSJOH
NBOVGBDUVSFSBOEBUUIFQIBSNBDFVUJDBMQBDLBHJOH UIFNPSOJOHTUBSUVQTJNQMZCFDBVTFFWFSZPOFLOFX
PQFSBUJPO5IFTFBDUJWJUJFTDPSSFDUFEUIFQSPCMFNBU UIBUUIFGJSTUTUBSUVQBMXBZTQSPEVDFEEFGFDUJWFT
UIFSPPUDBVTFBOEWFSJGJFETVDDFTTGVMDPSSFDUJPOBUUIF 8PSLBGGJSNFEUIBUTUBSUVQTDSBQXBTOPUJTPMBUFEUP
point of use. UIFNPSOJOH5IFTFDPOEJNQSPWFNFOUSFJOGPSDFE
5IFGPMMPXJOH$"1"DIBOHFTUPUIFQSPDFTTXFSF UIBURVBMJUZDBOOPUCFTBNQMFEPSJOTQFDUFEJOUPUIF
NBEFBUUIFDBSUPONBOVGBDUVSFS product.
tNew equipment operating procedure. The
TVQQMJFSJOWFTUFEJOBOBVUPNBUJDSFKFDUBSNUP POST CAPAVERIFYING AND MAINTAINING
iGMBHwUIFTUBSUPGBTIVUEPXOBTXFMMBTUIFFOEJOH VALIDATION AND PERFORMANCE
PGBTUBSUVQ$BQUVSFPGUIFEFGFDUTJNQSPWFEUIF &OIBODFENPOJUPSJOHPGUIFDBSUPONBOVGBDUVSFS
CMJTUFSQBDLBHJOHNBDIJOFFGGJDJFODZBOEZJFME BOEBEEJUJPOBMTBNQMJOHPGJODPNJOHDBSUPOTXFSF
tTraining.5IFNVUVBMJOWFTUNFOUJOTJUFBVEJUT JOTUJUVUFEUPDPOGJSNTVDDFTTGVMJNQMFNFOUBUJPOPG
XJUIPQFSBUPSMFWFMFNQMPZFFTPQFOFEUIFEPPSGPS $"1"BDUJWJUJFT#FDBVTFTVDDFTTGVMEFNPOTUSBUJPOPG
DBSUPOEFTJHOJNQSPWFNFOUTBTOFXFRVJQNFOU $"1"XBTQSPWFO SPVUJOFNPOJUPSJOHBOETBNQMJOH
was purchased. QMBOTXFSFBNFOEFEUPUBJMHBUFPSQSFEFMJWFSTBNQMFT
tNew sampling procedure.4BNQMJOHXBTOP GSPNUIFTVQQMJFS BNBKPSUJNFBOEMBCPSTBWFS
MPOHFSWJFXFEBTBSBOEPNJ[FEQVMMGSPNBMMPGUIF
DBSUPOT CVUSBUIFS BTBNFBOTUPDPOGJSNDPOUSPMT CONCLUSIONS
are working properly. "QSPCMFNJOWPMWJOHBTJNQMFQSPDFTTDBOPGUFOFYQPTF
tImpact on past manufacturing.1SFWJPVTMPUT GVOEBNFOUBMTBOELFZTUPFGGFDUJWFRVBMJUZDPOUSPMT*U
XFSFOPUDPOTJEFSFEUPCFBUSJTL5IFSFBTPOGPS BMTPDBOUFBDIMFTTPOTUIBUBSFOPUGPVOEJOUIFRVBMJUZ
UIJTJTUIBUDBSUPOTXFSFEBNBHFECFZPOEUIFQPJOU NBOVBM
PGCFJOHVTBCMFBOEDPVMEOPUCFTPME 5IJTQSPCMFNJOWPMWFEUIFVTFPGEFGFDUJWF
tValidation.#FDBVTFUIFQSPCMFNXBTEFUFSNJOFE TFDPOEBSZQBDLBHFT JF DBSUPOTDPOUBJOJOHUBCMFUTJO
UPCFBTFUVQBOEPQFSBUJPOBMQSPCMFN BOEOP CMJTUFSQBDLBHFT
BUBQIBSNBDFVUJDBMNBOVGBDUVSJOH
FRVJQNFOUPQFSBUJOHQBSBNFUFSTXFSFDIBOHFE QMBOU$BSUPOTXFSFGPVOEUPIBWFFJUIFSOPHMVF
WBMJEBUJPOXBTOPUOFDFTTBSZ)PXFWFS DPOUSPMT PSFYDFTTJWFBNPVOUTPGESJFEHMVF CPUIPGXIJDI
XFSFBEEFEUPDMPTFSBMJHONPOJUPSJOHEBUBUPUIF SFTVMUFEJONBDIJOFEBNBHFBOEQSPEVDUJPO
DSJUJDBMQPJOUTPGDPOUSPMBTUIFQSPDFTTXBTCVJMU EPXOUJNF5IFQSPCMFNXBTJOUFSNJUUFOUBOEBUBMPX
MFWFMEFGFDUSBUF CVUIBETJHOJGJDBOUSBNJGJDBUJPOT
5IFGPMMPXJOHDIBOHFTXFSFNBEFBUUIF MBSHFFYDFTTFTPGCBDLVQQBDLBHJOHNBUFSJBMT
QIBSNBDFVUJDBMQBDLBHFNBOVGBDUVSFS BEEJUJPOBMJODPNJOHNBUFSJBMJOTQFDUJPO BOE
Carton storage.*NQSPWFNFOUTJODBSUPOTUPSBHF TJHOJGJDBOUNBDIJOFEPXOUJNFJONBOVGBDUVSJOH
XFSFJNQMFNFOUFE-JNJUBUJPOTPOQBMMFUTUBDLJOHJO *OWFTUJHBUJPOBOEVMUJNBUFSFTPMVUJPOPGUIJTFWFOU
UIFXBSFIPVTFQSFWFOUFEFYDFTTJWFQSFTTVSFPOTUPSFE SFRVJSFEJOWPMWFNFOUPGTFWFSBMHSPVQTUISPVHIB
DBSUPOT5IJTFMJNJOBUFEUIFQPUFOUJBMGPSCPYCFOEJOH TVQQMJFSDFSUJGJDBUJPOJOJUJBUJWF"DPODFSUFEFGGPSU
BOEDVSMJOHUIBUDPVMEJNQBDUTFDPOEBSZQBDLBHF UPFTUBCMJTIBUSVTUJOHSFMBUJPOTIJQXJUIUIFDBSUPO
NBDIJOFPQFSBUJPO NBOVGBDUVSFSXBTJOJUJBUFE$MFBSPCKFDUJWFTXFSF
Verification of corrective action.8FUPPLB TUBUFEUPTPMWFUIFQSPCMFN5IFSFXBTOPNFOUJPOPG
OPWFMBQQSPBDIUPWFSJGZJOHUIFTVDDFTTPG$"1" BTTJHOJOHCMBNF5IJTNVUVBMVOEFSTUBOEJOHQSPWFE
BDUJWJUJFTBUUIFDBSUPONBOVGBDUVSFS5IFRVBMJUZVOJU UPCFUIFLFZUPPQFOBOEFGGFDUJWFEJBMPHVFPGJTTVFT
BVUIPSJ[FETIPSUUFSNEFQBSUVSFTGSPNGJSTUJOGJSTU 5IFGJSTUTFWFSBMNPOUITDPVMECFUFSNFEBQFSJPEPG
PVU '*'0
DPOUSPMTUPQSPWJEFJNNFEJBUFGFFECBDL iCVJMEJOHUSVTUwXIFSFPOMZTVTQJDJPOIBEQSFWJPVTMZ
UPUIFTVQQMJFS8JUIBMBSHFFYJTUJOHJOWFOUPSZ BOZ FYJTUFE.BOZPGUIFBTTVNQUJPOTUIFDVTUPNFSIFME
JNQSPWFNFOUTNBEFCZUIFTVQQMJFSXPVMESFRVJSFB UVSOFEPVUUPCFJODPSSFDU"OJNQPSUBOUMFTTPOXBT
XBJUPGBQQSPYJNBUFMZGPVSNPOUIT5IJTXBTUPPMPOH MFBSOFE5IFSFJTOPTVCTUJUVUFGPSDBSFGVMTUVEZPGB
UPEFUFSNJOFUIFFGGFDUJWFOFTTPGUIFQSPDFTTDIBOHFT QSPCMFN
(MVJOHEFGFDUTXFSFBMMCVUFMJNJOBUFE 6MUJNBUFMZUIFQSPCMFNSPPUDBVTFXBTEFUFSNJOFE
BOEBQQSPQSJBUF$"1"XFSFDPNQMFUFE5IFQSPCMFN
CARTON MANUFACTURER APPROACH TO SPPUDBVTFXBTUSJWJBMJNQSPQFSNBDIJOFRVBMJUZ
SAMPLING DPOUSPM 2$
BUTIVUEPXOBOETUBSUVQSFTVMUJOH
5IJTFWFOUVMUJNBUFMZSFTVMUFEJOBTJHOJGJDBOUMZ JOVOHMVFEPSPWFSHMVFEDBSUPOTCFJOHTFOUUPUIF
JNQSPWFEBQQSPBDIUPQSPDFTTTBNQMJOH5IFGJSTU DVTUPNFS5IFTJNQMFTPMVUJPOJOWPMWFEUXPBDUJPOTCZ

D. Wylie McVay, Jr.
UIFDBSUPONBOVGBDUVSFSSVOUIFQSPDFTTUPBMPHJDBM t$PNNPOTFOTFTIPVMECFUIFGJSTUJOWFTUJHBUJWF
TUPQQJOHQPJOUBOEUJHIUFS2$DPOUSPMPGDBSUPOTCFGPSF QBUIXBZQSJPSUPJNQMFNFOUBUJPOPGTUBOEBSE
BOEBGUFSBTIVUEPXOPSTUBSUVQ5IFTFDPNNPO QSPDFEVSFTBOETPQIJTUJDBUFENBUIFNBUJDT
sense corrections had significant process and financial 4UBOEBSEQSPDFEVSFTNBZOPUCFBQQSPQSJBUF
CFOFGJUTJONBOVGBDUVSJOHUISPVHIQVU SFEVDFEMBCPS HVJEBODFGPSQSPDFTTDPOUSPMSFMBUFEUSPVCMF
EFDSFBTFEXBTUF BOEEFDSFBTFENBUFSJBMFYQFOTFT shooting.
-FTTPOTMFBSOFEGSPNUIJTFWFOUXFSFBTGPMMPXT 5IFTFMFTTPOT XIJDIBSFVTVBMMZOPUGPVOEJOUIF
t5IFSFJTOPTVCTUJUVUFGPSDMFBSMZVOEFSTUBOEJOHB RVBMJUZNBOVBM NBZCFVTFGVMJOBEESFTTJOHBOZ
QSPCMFN&GGFDUJWFTPMVUJPOTTPNFUJNFTUBLFUJNF RVBMJUZQSPCMFNBOETIPVMECFJOUFHSBUFEJOUPUIF
BOEFGGPSUCVUQSPWJEFSFUVSOPOBTTFUCZIBWJOH DPNQBOZPSTJUFLOPXMFEHFCBTFGXP
CFUUFSQSPEVDUJWJUZ
t"USVTUJOHSFMBUJPOTIJQGBDJMJUBUFTQSPCMFNTPMWJOH ARTICLE ACRONYM LISTING
XJUIJOUIFDVTUPNFSPSHBOJ[BUJPO5SVTUJOH CAPA Corrective Actions and Preventive Actions
SFMBUJPOTIJQTBSFFWFONPSFJNQPSUBOU BOENPSF CGMP Current Good Manufacturing Practice
EJGGJDVMU XIFOUIJSEQBSUZWFOEPSTBSFJOWPMWFE FIFO First-In-First-Out
t"iGSFTIFZFTwBQQSPBDI JF MPPLJOHBUUIF PVC Polyvinylchloride
QSPCMFNCZBOFXBOEVOGBNJMJBSPCTFSWFS
NBZ QC Quality Control
RVJDLMZIFMQUPDMBSJGZUIFJTTVF5IFSFTIPVMECF
OPOFHBUJWFDPOTFRVFODFTXIFODBMMJOHJOOFX
QFPQMFUPBEESFTTBQSPCMFN
0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of GXP Compliance

Continuing Process Verification - 1

Uploaded by

Copyright:

Available Formats

Continuing Process Verification - 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Continuing Process Verification - 1

Uploaded by

Copyright:

Available Formats

CONTINUED PROCESS VERIFICATION

MONITORING AND MAINTAINING

Welcome to Maintaining Validation. KEY POINTS DISCUSSED

PROCESS VALIDATION Continued Process Verification 3

INTRODUCTION: WHAT IS t21 CFR 211.42-10(iv). i#VJMEJOHTBOE

4 PROCESS VALIDATION Continued Process Verification

tICH Q8 Pharmaceutical Development 

PROCESS VALIDATION Continued Process Verification 5

Table: Environmental sampling frequencies for DIBOHFTGPSUIFQBSBNFUFSCFJOHNPOJUPSFE PS

6 PROCESS VALIDATION Continued Process Verification

NPOJUPSJOHBMMPXTGPSPOHPJOHWFSJGJDBUJPOBOE '%" $'3 C

0SJHJOBMMZQVCMJTIFEJOUIF8JOUFSJTTVFPGThe Journal of Validation Technology

PROCESS VALIDATION Continued Process Verification 7

8 PROCESS VALIDATION Continued Process Verification

PROCESS VALIDATION Continued Process Verification 9

10 PROCESS VALIDATION Continued Process Verification

Figure 1: Prioritization for revalidation.

Discontinue production until revalidation is ready

SFRVJSFNFOUT FH MJOFDBQBDJUZ

PROCESS VALIDATION Continued Process Verification 11

Figure 2: When to revalidate. DIBOHFJTJNQMFNFOUFE5IJTXJMMFOTVSFOPUPOMZ

12 PROCESS VALIDATION Continued Process Verification

FWFOU)PXFWFS UIFDPOWFOUJPOIBTCFFOUPUSFBU Figure 3: Some means for detecting and reduc-

UBLFBNPSFTUSVDUVSFEBQQSPBDI UIFSFTIPVMECF SPC = Statistical Process Control PM = Planned Maintenance

PROCESS VALIDATION Continued Process Verification 13

14 PROCESS VALIDATION Continued Process Verification

THE BUSINESS MODEL PGFNQMPZJOHBDPOUSBDUPSJTUIFMPTTPGLOPXMFEHF

PROCESS VALIDATION Continued Process Verification 15

0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of Validation Technology

16 PROCESS VALIDATION Continued Process Verification

Dos and Donts of Control

Analysis and Control of Variation is dedicated B

PROCESS VALIDATION Continued Process Verification 17

18 PROCESS VALIDATION Continued Process Verification

Past due steps

PROCESS VALIDATION Continued Process Verification 19

mR chart for DC missed step delta

Figure 8: Control chart for production rate

X chart for shift B pen production per shift

0SJHJOBMMZQVCMJTIFEJOUIF8JOUFSJTTVFPGThe Journal of Validation Technology

20 PROCESS VALIDATION Continued Process Verification

The Dos and Donts of Control

Analysis and Control of Variation is dedicated INTRODUCTION

KEY POINTS DISCUSSED EXAMPLE FOUR

PROCESS VALIDATION Continued Process Verification 21

Figure 1: Conversion time monthly average.

X Chart for Conversion Time Monthly Average

mR Chart for Conversion Time Monthly Average

22 PROCESS VALIDATION Continued Process Verification

Figure 2: Product conversion time on a lot basis.

X Chart for Conversion Time by Lot

mR Chart for Conversion Time by Lot

t3FHJPOGSPNMPUUPMPU*UBQQFBSTUIBU EXAMPLE FIVE

PROCESS VALIDATION Continued Process Verification 23

24 PROCESS VALIDATION Continued Process Verification

Signal Moving Range of Metric

-PXFS$POUSPM$IBSU-JNJU-$- && BCTPMVUFEJGGFSFODFCFUXFFOUIFUXPEBUBQPJOUT

PROCESS VALIDATION Continued Process Verification 25

"TBTJEFOPUF UIFPUIFSJTTVFXJUIUIF.#" PVSQSPDFTT UIFOUIFSFDBOCFHSFBUBEWBOUBHFTJO

tICH Q8 Pharmaceutical Development

NPOJUPSJOHBMMPXTGPSPOHPJOHWFSJGJDBUJPOBOE '%" $'3 C

0SJHJOBMMZQVCMJTIFEJOUIF8JOUFSJTTVFPGThe Journal of Validation Technology

SFRVJSFNFOUT FH MJOFDBQBDJUZ

Figure 2: When to revalidate. DIBOHFJTJNQMFNFOUFE5IJTXJMMFOTVSFOPUPOMZ

FWFOU)PXFWFS UIFDPOWFOUJPOIBTCFFOUPUSFBU Figure 3: Some means for detecting and reduc-

0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of Validation Technology

0SJHJOBMMZQVCMJTIFEJOUIF8JOUFSJTTVFPGThe Journal of Validation Technology

t3FHJPOGSPNMPUUPMPU*UBQQFBSTUIBU EXAMPLE FIVE

-PXFS$POUSPM$IBSU-JNJU-$- && BCTPMVUFEJGGFSFODFCFUXFFOUIFUXPEBUBQPJOUT

0SJHJOBMMZQVCMJTIFEJOUIF"VUVNOJTTVFPGThe Journal of Validation Technology

Figure 5: Distribution curve for USP 30 content uniformity JTUIFTBNQMFTJ[F

1.33%, 16.89% GPMMPXBCJOPNJBMEJTUSJCVUJPO*OB

Figure 8: Distribution of nonconforming rates in samples "TJMMVTUSBUFEJO'JHVSF BHSFBU

40.0% Mean = p = 0.01 = 1% 2$UFTUT

GSPNUIFCBUDI BOEUIBUUIFEJTUSJCVUJPOPGUBCMFU PG2WBMVF5IFQSPCBCJMJUZDIBSUTGPSEFUFSNJOJOH

JODMVEJOHTQFDJGJDBUJPOTw5IFIJHIDPOGPSNBODF +4#FSHVNBOE)-J i"DDFQUBODF-JNJUTGPSUIF/FX

0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of Validation Technology

VOUJMUIFQSPDFTTVOEFSFYBNJOBUJPOJTLOPXOUP 8"4IFXIBSU Economic Control of Quality of Manufactured

0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of Validation Technology

&WFSZEBZEFWJBUJPOTBSFSBJTFEBOEJOWFTUJHBUJPOT BSFSBSF*OBDDPSEBODFXJUI+VSBOTSVMF BCPVU

0SJHJOBMMZQVCMJTIFEJOUIF4QSJOHJTTVFPGThe Journal of Validation Technology

WBMJEBUJPOSFQPSUEJTDVTTFTSFTVMUTBOEQSPWJEFT Validation plan."GPSNBMSFRVFTUUPJOJUJBUF