Reprinted from PHARMACEUTICAL ENGINEERING · development base, API, drug products, and parenteral...

Reprinted from

PHARMACEUTICAL ENGINEERING®

The Official Journal of ISPE

July/August 2003, Vol. 23 No. 4

Country Profile - Singapore

2 PHARMACEUTICAL ENGINEERING JULY/AUGUST 2003 ©Copyright ISPE 2003

This newfeature inPharmaceuticalEngineering isdesigned sothat you cantear it out, threehole drill(if desired),and keep it withother CountryProfiles as theyare published.

Look for theCountry Profileon Belgium inthe September/October issue ofPharmaceuticalEngineering.

Dear ISPE Members and Readers,

It is with great pleasure that I present the SingaporeCountry Profile in this issue of Pharmaceutical Engineering.I am Italian, and as probably many of you know, Italiansare very proud of the natural and artistic beautyof our country. However, even with such high standards, I have fallenin love with Singapore's tropical climate, green vegetation, hightechnology, culture, and people since my plane landed at Changi airportmore than two years ago. Although the most endearing attribute to thiscity-state is the people who are always keen to smile, try to help youeven for the most trivial things, and have a great capacity for learning.It is with this personal joy and conviction of the merits of Singapore thatI introduce you to this short, but hopefully informative and interestingmicrocosm of the world of pharmaceuticals in 2003 and beyond inSingapore.

Singapore is a young, vibrant country formed by a friendly split withMalaysia in 1965. Its primary assets are its location, harbor, stableclimate, and people who at around 4.5 million (more than Ireland) havetransformed it over this period into the worlds busiest harbor, largestrefinery area, and a top ten global economy. This is all based primarilyon manufacturing and the process industries; and the bio-sciences havebeen targeted as one of the high technology intellectual propertysectors for future growth.

While the first multi-national pharmaceutical company manufacturingsemi-synthetic antibiotics came here in the early 1970s, our manufac-turing industry has significantly grown in the last 5-7 years. Singapore,with a pharmaceutical plant investment of approximately $3 billion, isprobably the most dynamic and innovative biomedical hub in Asiaoutside Japan. Activities range from a growing basic research anddevelopment base, API, drug products, and parenteral biotech proteinsmanufacture to a large regional center for clinical trials.

Its Economic Development Board is actively selling around the worldSingapore as an integrated, strategic global bio-science location. It isa politically stable, relatively low cost environment with an ethicallystrong business and government culture and hard working people. Ithas an equal or greater growing population of skilled staff (both forproduction and service industry) as any other equivalent pharmaceuti-cal center, to meet the expected growth in new plants, and sufficientnewly created land to support a trebling of the current plants at least.The Government Financial packages are more than attractive.

I hope you will find this Singapore Country Profile interesting enoughto put Singapore in your sights when the subject of a new pharmaceu-tical development location is discussed. Who knows, a cold tiger beer,a bowl of tasty laksa, and the ancient culture of the Asian continentawait you once your facility is completed and operational, together witha sound financial return on your investment!

Yours Truly,Dr Ing Roberto GardellinChairman, ISPE Singapore Affiliate


JULY/AUGUST 2003 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2003

Singapore - The Premier Biomedical Hubof Asiaby the ISPE Singapore Affiliate

Singapore has enjoyed phenomenal growth overthe last four decades despite its small size andpopulation - just 4.5 million people - and lackof natural resources. Its per capita Gross Do-

mestic Product (GDP) has been growing at an averageannual rate of 10.5 %, swelling from $512 in 1965 tonearly $21,000 in 2002, and now trails behind onlyJapan and Hong Kong for the highest per capitaincome in Asia.

Despite its historical importance during its days as aBritish crown colony as a strategically located tradingport linking the West with the East, Singapore’s mi-raculous economic growth has been achieved by focus-ing on manufacturing and productivity. It built ship-yards, attracted global oil refineries to build on re-claimed land, and later followed the technology waveonto manufacturing and now designs of disk drivesand semiconductors.

Today, manufacturing and services are the twin eco-nomic engines of growth, with the chemicals, electron-ics, and the engineering clusters as the key pillars ofSingapore’s economy.

To diversify and strengthen Singapore’s economic re-silience, the government has been aggressively devel-oping the Biomedical Sciences cluster as another key

pillar of the country’s economy. Since June 2000, thisinitiative is jointly driven by the Biomedical SciencesGroup (BMSG) of the Singapore Economic Develop-ment Board (EDB) and the Biomedical Research Coun-cil (BMRC) of the Agency for Science, Technology, andResearch (A*STAR). EDB is the Government bodyresponsible for industrial development while A*STARfunds, coordinates, and directs public research, as wellas promotes public awareness of science and technol-ogy in Singapore.

In addition, an International Advisory Council com-prising pre-eminent scientists from the US, Europe,and Australia advises the government on various bio-medical initiatives covering R&D, industry develop-ment, education, and healthcare. The Council is Chairedby Sir Richard Sykes of Imperial College in Londonand Co-Chaired by Dr Sydney Brenner of the SalkInstitute for Biological Studies in California.

A Timely Boost To the EconomyAs the country emerges from one of the worst reces-sions since its independence, the solid performance ofthe pharmaceutical industry has provided a timelyboost to the economy. Measured on a year-on-yearbasis, the sector has shown strong and consistentgrowth in total manufacturing output, jobs creation,and value-added for the past five years.

Merlion overlooking Singapore River.

Singapore Fact Sheet (2002):

Physical FactsPopulation: 4.5 millionLand Area: 682.3 sq kmAverage Daily Temperature: 26.8 - 31°CAnnual Rainfall: 2345 mm

EconomyCurrency: Singapore Dollars

(S$1.79:US$1)GDP: US$87 billion/S$155.7 billionPer Capita GDP: US$20920

Unemployment 4.2%



Manufacturing output from the sector demonstrated58 % growth last year providing a bright spark amidstthe global slowdown in the electronics industry. LastDecember saw a 131% increase in pharmaceuticalexports as new plants were set-up to produce higher-value drugs. This helped the Singapore economy reg-ister positive growth, reversing the recession experi-enced in 2001. And, in a year of overall rising unem-ployment, the industry created about 1,040 jobs, rep-resenting a 136% increase over the previous year.

The pharmaceutical industry in Singapore may havehad its beginnings all the way back in the 1970s, butthe past five years have seen it being thrust into thelimelight as the island-state seeks to become a globalBiomedical Sciences hub in Asia. Today, the pharma-ceuticals sector is a $4.4 billion industry in Singaporewith a strong critical mass of leading internationalcompanies.

Quality manpower, good infrastructure, global marketnetworks, and strong intellectual property rights pro-tection have led companies like Aventis,GlaxoSmithKline (GSK), Merck Sharp & Dohme,Schering-Plough, Pfizer, and Wyeth to invest morethan $2.4 billion in manufacturing facilities here toproduce Active Pharmaceutical Ingredients (APIs) andfinished products for global markets. With a steadypipeline of manufacturing investments by leading phar-maceutical players, Singapore is rapidly growing fromstrength to strength in its status as a key manufactur-ing launch pad for the global pharmaceutical market.

For instance, Schering-Plough officially opened itsnew $100 million Biotech Sterile Manufacturing Facil-ity and $78 million Tablet Facility last November.These two facilities will complement its internationalproduct manufacturing operations to meet increasingglobal demand. Schering-Plough also announced that

it would invest a further $200million to build a third multi-purpose plant with commer-cial production from 2005. Withthis third plant, Schering-Plough’stotal investment commitments inSingapore would exceed $1 billionand total staff strength would reach800.

The Manufacturing Hub:Tuas Biomedical Park

The majority of the industry’s manu-facturing plants are located at thewestern end of the island at a spe-cially designated area called the TuasBiomedical Park. Its origins may be

traced back to the mid-1990s when the Singaporegovernment allocated a 50 hectare site at Tuas, theindustrial hub of the country, as a pharmaceuticalmanufacturing zone, known then as the Pharma Zone.

It has to date proved successful in establishing acluster of leading foreign companies such as Pfizer,Wyeth, and Merck, and has expanded into 170 hect-ares of prepared land for the further clustering ofpharmaceutical and biotech manufacturing operationsand other shared services.

Excellent infrastructures of roads, sewer lines, anddrainage systems have been put in place. There also arefuture initiatives to provide third party services such asa centralized waste treatment and utility plant for thepharmaceutical facilities there. Due to strong demand,an additional 150 hectares has been developed.

Ingredients for Pharmaceutical Boost -Right Infrastructure and Capabilities

While the tremendous progress of the pharmaceuticalindustry in Singapore may appear to be an overnightsuccess, in reality Singapore has devoted considerableeffort toward developing the right infrastructure andcapabilities for this industry to flourish. These includecreating plug and play environments for R&D andmanufacturing activities as well as providing a strate-gic mix of financial incentives and grants for R&D andmanpower training to help companies jumpstart theiroperations.

Indeed, the pharmaceutical industry in Singapore hasa 30-year history - the first players actually arriving onthe island as far back as 1973. In that year, SmithKlinestarted the ball rolling with its antibiotics plant. Glaxothen started its operations in 1979 when it built itsfirst active ingredients plant – a $150 million chemi-cals plant.

Changi Airport control tower.



While both companies raised their fixed asset invest-ments in Singapore, they remained very much the onlyplayers in Singapore for another 14 years before Fisonsopened its chemical plant in 1993.

And in 1997, US-based Schering-Plough set up a $260million multipurpose chemical plant, which openedthe flood-gates for many other US drug companies tobring in their manufacturing operations to Singapore.

Beyond Manufacturing: R&DIn an industry where Research and Development (R&D)is critical to long-term sustainability, many compa-nies, such as Novartis, Eli Lilly, and Bristol-MyersSquibb have invested in R&D activities ranging frombasic research to clinical development, attracted bySingapore’s multi-ethnic population, well-developedclinical and regulatory infrastructure, strong IP frame-work, easy access to regional patients as well as strictadherence to international clinical standards.

Some, like GSK and Schering-Plough, have augmentedtheir manufacturing operations in Singapore by build-ing capabilities in process development. Schering-Plough’s Chemical R&D Center, which was recentlycompleted, will carry out process development andprocess optimization activities.

In addition, pharmaceutical companies have broad-ened the scope of their clinical research activities inSingapore to include early phase trials. In March 2001,Pharmacia established a 24-bed clinical pharmacol-ogy center at Singapore General Hospital. It is thesecond company after Eli Lilly to invest in clinicalpharmacology facilities with Singapore’s hospitals.

Eli Lilly also launched its corporate R&D center lastyear dedicated to systems biology research. This is thefirst major commitment made by any pharmaceuticalcompany in the field of systems biology with thespecific purpose of accelerating the drug discoveryprocess. With a $140 million R&D budget over fiveyears, Eli Lilly will employ approximately 50 scien-tists and information technology professionals.

And taking advantage of Singapore’s tropical location,Novartis is setting up a research center to find newdrugs and treatments for tropical diseases, initiallyfocusing on tuberculosis and dengue fever.

Apart from the multi-national firms, local firms alsohave made significant progress in 2002.One such ex-ample is home-grown biotechnology firm - MerLionPharmaceuticals, formed only in July last year, whichhas successfully raised $13.5 million in equity fundingdespite a tough global financing environment, andsecured collaborations with Abbott Laboratories,

Tuas Biomedical Park is located at the west end of Singapore.



Athelas, Fujisawa, Genome Therapeutics, KuDOSPharmaceuticals, and Merck & Co. Boasting one of theworld’s largest and most diverse natural productslibraries derived from bacterial, fungal, plant, andmarine organisms, the company is looking for activemolecules from natural products that work againstdiseases like cancer and diabetes.

R&D City: BiopolisSingapore is building a dedicated biomedical researchpark, known as “Biopolis,” to house BMRC’s five bio-medical research institutions as well as R&D laborato-ries of pharmaceutical and biotechnology companies.Targeted to start operating from mid 2003 onward, the2 million square foot R&D complex will incorporatefacilities tailored for the Biomedical Sciences, includ-ing laboratory space for private biomedical companies,incubators to nurture start-up companies, animal han-dling facilities, as well as laboratory support services.Central facilities such as shared R&D facilities, audi-torium, and lifestyle amenities also will be easilyaccessible and available to the tenants at Biopolis.

Located near the National University of Singapore,National University Hospital, and the Singapore Sci-ence Parks, Biopolis aims to be a breeding ground forsynergy and collaboration of new research discoveriesbetween the public and private sector researchers.

Mr Philip Yeo, Chairman of A*STAR as well as Co-Chairman of EDB, highlighted: “Biopolis will be avibrant community of Human, Intellectual, and Indus-trial Capital, with leading scientists and top-rate orga-nizations from all parts of the world congregating at afocal point for cutting-edge research. Researchers willbe able to interact and exchange ideas, collaborate,

and leverage on the different strengths avail-able in both the public research institutes andthe private companies.”

The Future and Getting ThereMoving forward, Singapore is on track to achieving itstarget of $6.7 billion in Biomedical Sciences manufac-turing output by 2005, according to Philip Yeo.

And not only will output continue to grow in volume,the range of activities undertaken by the industry isexpected to expand further in terms of the breadth anddepth of its manufacturing base. In particular, biologicsmanufacturing - the large-scale production of protein-based drugs - is likely to increase its presence here assuch drugs are expected to account for 50 to 60% of newdrugs in the future.

A-Bio Pharmaceuticals, a start-up contract biologicsmanufacturer will target leading pharmaceutical andbiotechnology companies to provide contract manufac-turing services, specializing in mammalian cell cul-ture. Its proposed manufacturing plant in the TuasBiomedical Park is expected to be ready by 2007.

Building Human CapitalTo meet the growing need for skills and knowledge inboth the manufacturing and R&D segments of theindustry, Singapore is trying to attract scientific andtechnical expertise from around the globe. In additionto the country being a cosmopolitan place to live - withits high public safety, cleanliness, excellent publictransport, and English-speaking community - researchfellowships and grants are being provided to special-ists in the biomedical field.

R&D City: Biopolis.

Singapore Fact Sheet (2002):

Living in SingaporeLiteracy Rate: 93.7%Life Expectancy: 78.7 yearsHome Ownership: 93.6%Population Density: 6055 per sq km

Official Languages:English (for administration), Chinese,Malay, Tamil

Ethnic Composition:Chinese: 76.5%Malays: 13.8%Indians: 8.1%Others: 1.6%



At the same time, Singapore is aggressively buildingits own pool of local talent by strengthening the cur-riculum for the Biomedical Sciences at all levels.

The Ministry of Education has modified the country’sprimary and secondary school curriculum to providefoundational understanding of the Biomedical Sci-ences as well as basic training in modern scientificinvestigative skills.

At the post-secondary level, all four of Singapore’spolytechnics, which provide tertiary-level vocationaltraining, are now offering courses for biomedical labtechnicians and research assistants. And at the uni-versity level, the National University of Singaporerevamped its life sciences curriculum last year to putgreater emphasis on research. The country’s otheruniversity, the Nanyang Technological University,has recently established its School of Biological Sci-ences and had its first intake of 100 students in Julylast year.

Various scholarships for both undergraduate and post-graduate degrees also have been set up. A*STAR’sNational Science Scholarships, which was launched in2001, will support undergraduate and postgraduatetraining up to PhD and postdoctoral level for about 600research scientists in the Biomedical Sciences.

And to ensure that training is not restricted totheory, the EDB also has developed an indus-trial training program for biopharmaceuticalsmanufacturing. Under the Training and At-

tachment Programme (TAP), engineers and scien-tists will be sent for a period of between 12 and 18months to leading companies in Europe and US where

they will be trained in the areas of process develop-ment, validation, and quality assurance.

Protecting Intellectual PropertyTo further encourage researchers to create intellectualproperty on the island, the Intellectual Property Officeof Singapore was launched in 2001 to provide theinfrastructure, platform, and environment for greatercreation, protection, and exploitation of intellectualproperty.

Singapore has achieved full compliance with the WorldTrade Organization’s Trade-Related Aspects of Intel-lectual Property Rights Agreement one year before the2000 deadline. It also has been ranked by the Politicaland Economic Risk Consultancy as having the bestintellectual property rights protection in Asia since1997. Singapore is also a signatory to the World Intel-lectual Property Organization, the Paris Convention,the Budapest Treaty, and the Patent CooperationTreaty. The Health Sciences Authority underSingapore’s Ministry of Health also provides a compre-hensive regulatory framework for the evaluation andmarketing approval of all therapeutic products.

On TargetSince the launch of the Biomedical Sciences initiativein June 2000, Singapore has successfully built a grow-ing reputation in the international biomedical commu-nity for its comprehensive plans, stringent IP protec-tion, and strong commitment to develop the Biomedi-cal Sciences cluster. Its success in both manufacturingand R&D are a clear signal that the island-state hasthe right mix of public research and industry involve-ment for the high value-added and technology-inten-sive Biomedical Sciences cluster.



Singapore’s Regulatoryand IndustryDevelopment

In the 1960s, the pharmaceuti-cal manufacturing industry inSingapore comprised mainlythe local generic manufactur-

ers and Singapore had no regula-tory GMP audit program then. Itwas not until 1973 that BeechamPharmaceuticals, a UK-based com-pany, became the first Multi-Na-tional Company (MNC) manufac-turer to set up a plant in Singaporeto manufacture bulk semi-syn-thetic penicillins.

Today, there are at least 10 MNCpharmaceutical manufacturingfacilities in Singapore, includingSchering-Plough, GlaxoSmithKline, Aventis-Pharma, WyethPharmaceuticals, Merck, Pfizer,and Baxter Healthcare.

Following the closure of the Gov-ernment Production Laboratoriesin 1986, the licensing of pharma-ceutical manufacturers and theregistration of medicinal productscommenced the following yearunder the framework of the Medi-cines Act.

A GMP Unit was established in1997 within the Ministry of Health(MOH) to deal with the increasingtypes and number of manufactur-ers, and to manage the increasingspecialization in the field of GMP.The licensing of Chinese Propri-etary Medicine (CPM) also beganin 1999.

More recently, on 1 April 2001, theHealth Sciences Authority (HSA)was established as a statutoryboard of MOH. HSA comprises

eight professional centers, includ-ing the Centre for PharmaceuticalAdministration (CPA), which ad-ministers the regulation of drugsand health-related products. Theother professional centers includethe Centre for Analytical Science(CAS), Centre for Drug Evalua-tion (CDE) Centre for ForensicMedicine (CFM), Centre for Fo-rensic Science (CFS), Centre forMedical Device Regulation(CMDR), Centre for Radiation Pro-tection (CRP), and the Centre forTransfusion Medicine (CTM).

CPA has four divisions, namelythe Manufacturing and QualityAudit (upgraded from GMP Unit),the Product Evaluation and Reg-istration, the Compliance andComplementary Medicine, and thePharmacovigilance, Communica-tions and Research Divisions.

The Manufacturing and QualityAudit arm of CPA comprises threeUnits, namely the GMP Audit Unit,the GDP Audit Unit, and the Cer-tification Unit.

• The principal functions of the

GMP Unit include the audit andlicensing of manufacturers ofsterile and non-sterile medici-nal products, CPM, cosmetics,as well as CLS.

• The principal function of theGDP Unit is the audit and li-censing of importers, wholesaledealers, and the retail and hos-pital pharmacies.

• The Certification Unit processesand grants various certificatessuch as the Certificate of a Phar-maceutical Product and Certifi-cate of Licensing Status (underthe WHO CertificationScheme), the Free Sales Cer-tificate and other Export Cer-tificates, as well as GMP cer-tificates

Accession of Singaporeto the Pharmaceutical

Inspection Co-operationScheme (PIC/S)

In line with the national goal ofSingapore to be a life sciences hub,the GMP Unit embarked on a qual-ity journey to benchmark itselfagainst overseas centers of excel-lence in the field of GMP audit andlicensing of pharmaceutical manu-

The Regulatory System of Singaporeby Dr. Clarence Tan, Chief Executive Officer,Health Sciences Authority

Singapore Health Sciences Authority building.



facturers. In July 1997, a formalapplication to accede to PIC/S wassubmitted.

PIC/S comprise countries withequivalent high standards of GMPinspection system, and include theEuropean Union countries, Swit-zerland, Australia, and Canada.Two PIC/S delegations visitedSingapore in April 1999 and No-vember 1999 respectively to assessits system of GMP inspection andlicensing of pharmaceutical manu-facturers. The PIC/S delegationsconcluded that the Singapore sys-tem of GMP inspection and licens-ing can now be considered to be“equivalent to that of PIC/S mem-ber authorities, and Singapore hasset a benchmark for other GMPinspectorates in the region tomatch.” With effect from 1 January2000, Singapore became the firstAsian country to accede to PIC/S.

With HSA’s membership of PIC/S,Singapore is now in a position topursue Mutual Recognition Agree-ments (MRAs) with other PIC/Scountries, beginning with Austra-lia. An MRA on GMP Inspectionwas signed between the Govern-ments of Singapore and Australiaon 26 February 2001. The signingof this MRA means that the Thera-peutic Goods Administration(TGA) of Australia now acceptsthe GMP audit reports and theconclusions of the GMP Auditorsof HSA and vice-versa.

Singapore and Japan also havesigned an Economic PartnershipAgreement on 13 January 2002,which also included a Joint State-ment on Pharmaceutical GMP In-spection, which provides for theexchange of GMP audit reports

between Singapore HSA andthe Japanese Ministry ofHealth, Labour, and Wel-fare (MHLW).

With these developments, the sta-tus of Singapore as a regional life

sciences and pharmaceutical hubhas been enhanced considerably.

The GMP Audit Systemof Singapore

The GMP inspection system ofSingapore follows closely the in-ternational practice of PIC/S. Therisk assessment approach for de-termining the frequency of GMPaudits is used.

For finished dosage forms,Singapore has adopted the PIC/SGMP Guide for Medicinal Prod-ucts as its GMP standard. In thecase of Active Pharmaceutical In-gredients (APIs), the PIC/S GMPGuide for API (equivalent to theInternational Conference on Har-monization Q7A Guidelines) hasbeen adopted by Singapore as thecorresponding standard.

Licensing ofPharmaceutical

Manufacturers inSingapore

The Medicines Act of Singaporestates that no person shall manu-facture or assemble any registeredmedicinal product unless he has amanufacturer’s license, and the li-censing authority shall take intoconsideration the following crite-ria before granting a Manufac-turer’s License:

• proposed manufacturing opera-tions

• details of the premises• equipment used for manufac-

turing and QC• qualifications of key personnel• Security of the premises and

the maintenance of adequatewritten procedures and records

Standard Provisionsfor a Manufacturer’s

LicenseThe standard conditions or provi-sions for a manufacturer’s licenseare set out in the Fourth Scheduleto the Medicines (Licensing, Stan-dard Provisions, and Fees) Regu-

lations. The holder of aManufacturer’s License must com-ply with the specified provisions,which includes complying with thePIC/S Guide to GMP for MedicinalProducts, as revised or amendedfrom time to time.

Strategic Directionsand Challenges

HSA will continue with its domes-tic program to internationalize thelocal pharmaceutical manufactur-ing industry standards, in particu-lar the generic product suppliersand the manufacturers of CPM.

Seminars and workshops will con-tinue to be organized on qualitytopics. The risk-based GMP auditprogram will continue, and wherenecessary, the appropriate regu-latory actions taken against recal-citrant companies and non-con-forming manufacturers.

In the near future, HSA also willhave to pay attention to new cat-egories of products such as bio-technology products, new types ofAPIs, clinical trial products, andhealth supplements. It is verylikely that HSA will participatemore actively in international har-monization of GMP standards andaudit systems through PIC/S.

More bilateral government-to-gov-ernment MRAs on GMP inspec-tion, which will result in greaterindustry benefits, are expected tobe negotiated and signed. An over-seas GMP audit program involv-ing more than 1000 overseasmanufacturers also is expected tobe implemented soon.

Questions relating to thearticle may be directed to:Manufacturing and Quality

Audit DivisionCentre for Pharmaceutical

AdministrationHealth Sciences Authority

SingaporeEmail: [email protected]

Web site: www.hsa.gov.sg



The Early Days

Dr. Miranda Yap is a Direc-tor of one of Singapore’spublic research institutes- the Bioprocessing Tech-

nology Centre (BTC) - and is con-sidered by many to be a pioneer inthe nation’s burgeoning biomedi-cal sciences research efforts. Herinterest in bioprocess science andtechnology was kindled while shewas pursuing a Master’s degree inbiochemical engineering at Uni-versity College London in 1973,after earning her basic degree inapplied chemistry from the Na-tional University of Singapore(NUS), then known as the Univer-sity of Singapore.

Following her Master’s degree, shejoined Singapore Petroleum Com-pany, a local oil refinery, as achemical engineer just as many ofher peers were working for the oiland gas industry, which was boom-ing in Singapore during the 1970s.

Recalls Dr. Yap: “Many of my un-dergraduate classmates workedfor petroleum companies and uponretirement, were given greatgolden handshakes! But candidly,I found such a ‘regular’ job ratherstifling.”

And so in 1975, she embarked on adoctorate program in chemicalengineering at the University ofToronto. After earning her PhD infour years, Dr. Yap spent anotherthree years in the United Statesconducting postdoctoral research.But even as career opportunitiesabound for her in the North Ameri-

can continent, her heart yearnedfor home and her family. In 1982,she returned to Singapore andjoined NUS.

“I recall that as an undergraduateI was always harboring thoughtsof a teaching career in NUS,” shesays, “So in 1982, my husband andI made a conscious choice to returnto Singapore.”

Fortunately for her, the early 1980swere a good time for academics toreturn to Singapore. Researchmoney was readily available asthe Government was then seekingto build up the nation’s local uni-versities. In addition to ample

funding, researchers also weregiven relatively free reign in defin-ing their research areas, she says.But even with the early support ofthe Government, Dr. Yap’s jour-ney into research was not an en-tirely smooth one.

BioprocessingTechnology Centre (BTC)and Manpower Training

As Dr. Yap continued in her aca-demic career over the next fewyears, the government soon iden-tified a need to initiate manpowertraining in the area of bioprocesstechnology as it sought to attractand anchor biopharmaceuticalcompanies to Singapore in a bid to

A personal perspective from Dr. Miranda Yap,Director of Bioprocessing Technology Centre, Professor ofNational University of Singapore (NUS), Department ofChemical and Environmental Engineering

R&D In Singapore: A New Challenge

Bioprocessing Technology Centre Building (BTC).



expand the island’s manufactur-ing base.

A task force comprising of NUSfaculty from various departmentsand industry representatives wasset up to look into establishing acenter, which would complementthe upstream activities of the In-stitute of Molecular and Cell Biol-ogy. The IMCB had been set up in1987 and was Singapore’s firstpublic biomedical sciences re-search institute, focusing on basicresearch in molecular genetics,including cell regulation, cell cyclecontrol, and genomics.

Being a member of the task force,Dr. Yap was instrumental to theformation of the BioprocessingTechnology Unit (BTU). Her pro-posal to initiate the unit in NUS’Chemical Engineering departmentwas accepted in 1990 and the BTUwas born with a $6 million grantfrom the Singapore Governmentand 3,000 sq feet of laboratoryfloor space. The BTU was subse-quently renamed the BTC and tookon the additional role of a nationalR&D center, with funding fromSingapore’s National Science andTechnology Board.

In starting up the BTC, the mainchallenge that Dr. Yap faced wasthe sourcing for experienced se-

nior scientists. At that time,bioprocessing technologywas still a nascent re-search area in Singapore,

and so it was difficult to at-tract the right people then, sherecalls. Nonetheless, she eventu-

ally managed to find enough se-nior scientists from overseas andbright local graduates also helpedto fill places as they became at-tracted to new career choices inthe biopharmaceutical sector.

Today, the BTC continues to playa pivotal role in manpower train-ing with core strengths in expres-sion engineering, animal cell cul-ture, downstream purification, andanalytics focused on enhancingproduct yields and quality.

Biomedical ResearchCouncil (BMRC)

Together with the IMCB and threeother research institutes - the Ge-nome Institute of Singapore, theBioInformatics Institute, and theInstitute of Bioengineering andNanotechnology, the BTC is one ofthe five pillars of Singapore’s Bio-medical Research Council (BMRC),which was established in October2000 to coordinate and support bio-medical research in the public sec-tor. Apart from funding public bio-medical research initiatives, theBMRC’s other role is to build up atalented pool of biomedical research-ers in Singapore. It has establishedseveral manpower development ini-tiatives that include scholarshipsand exchange programs.

Next up, the BMRC will be movingits member research institutes intoa new dedicated biomedical re-search park called “Biopolis.” The194 hectare research center, whichis expected to be completed in June2003, will house all of BMRC’sR&D activities from basic drug

discovery research to clinical de-velopment to medical devices re-search. Sited near NUS, hospitalsand other research institutes, theBMRC research arms hoped to seeda vibrant research community byattracting private industry re-search from both multi-nationaldrug firms and local biotechnologystart-up companies.

Indeed, the story of biomedical sci-ences research in Singapore wouldnot be complete without acknowl-edging the growing number of lead-ing pharmaceutical companies,which have set up R&D operationson the island.

For example, Eli Lilly recently es-tablished its state-of-the-art cen-ter for systems biology, its firstoutside of the US to look into thedevelopment of computationaltools for drug discovery.

Local start-up companies also havenot been left out of the fray. Theseinclude ES Cell International - astem cell company - that arosefrom research done at NUS,Monash Institute of Reproductionin Australia and Development andHadassit Medical Research Ser-vices and Development in Israel.

As Dr. Yap aptly sums it up, “Re-search is more than just a passingfancy, but is a life line forSingapore. This is because there isa great need to couple manufac-turing and R&D in high knowl-edge-based industries.

“The key attractions for such com-panies to locate here for manufac-turing are the availability of ahighly qualified manpower pooland relevant technologies whichwill add value to the industry.

“Thus, it is critical to meet theseneeds through the research insti-tutes and universities to build uprelevant manpower and technolo-gies to attract them.”

Institute of Molecular and Cell Biology Building (IMCB).



Singapore is positioned as the choice locationfor global manufacturing, supply chain man-agement, distribution, as well as upstreamactivities including process development, clini-

cal development, and R&D for Biomedical Sciences. Inclose partnership with industry, EDB’s BiomedicalSciences Group (BMSG) focuses on broadening anddiversifying the range of Biomedical Sciences activi-ties in Singapore and ensuring that a sound support-ing infrastructure is in place.

Over the years, Singapore’s base of pharmaceuticalmanufacturing activities has expanded from primarymanufacturing by companies such as Aventis andGlaxoSmithKline to include secondary manufacturingsuch as tabletting, formulation and finishing by Merckand Wyeth, and nutritionals manufacturing by Wyeth.Schering-Plough also has added biotechnology lyo-philization into Singapore’s host of high value-addedmanufacturing activities. Singapore’s ability to exten-sively support production and manage the supplychain of high-value pharmaceutical products to theglobal markets through its excellent infrastructure,strong IP framework, and availability of skilled man-power has been strengthened through the breadth anddepth of such manufacturing activities.

Promising growth areas such as the biopharmaceuticalssector will continue to be nurtured as Singapore ex-tends its BMS industry capabilities. The recent open-ing of the small-scale $19 million cGMP facility of theBiopharmaceutical Manufacturing Technology Cen-tre will introduce clinical-grade biologic manufactur-ing capabilities for the production of monoclonal anti-bodies and other biopharmaceuticals.

Singapore is also an attractive and strategic locationfor companies to conduct and manage clinical develop-ment activities in Asia. Pharmaceutical companies

like Bristol-Myers Squibb, GlaxoSmithKline, EliLilly, Merck Sharp and Dohme, and NovoNordisk base their clinical development teamsin Singapore to oversee clinical trials in the

region. Success factors include Singapore’s multi-ethnic population, well-developed clinical and regula-tory infrastructure, easy access to regional patients, as

well as strict adherence to international clinical stan-dards. These companies work closely with the localhospitals and Contract Research Organizations (CROs)such as Covance, Icon, and Quintiles to conduct earlyto late stage trials in Singapore and in the region. Suchdevelopments are further bolstered on the regulatoryfront, where the Health Sciences Authority (HSA)under Singapore’s Ministry of Health ensures thatthere is a strong regulatory framework that is support-ive of clinical research.

Apart from manufacturing and clinical development,the Singapore government has been nurturing thegrowth of a critical mass of companies undertakingR&D in Singapore. In 2000, a $600 million fund wasintroduced to encourage companies to establish theirR&D centers or spin-off research projects. This hasattracted large pharmaceutical and smaller biotech-nology companies to undertake R&D in Singapore,including Novartis, Eli Lilly, Pharmalogicals Research(a joint-venture between Chugai Pharmaceuticals andMitsui & Co), and Agenica.

To further facili-tate the growth ofBiomedical Sci-ences companieswith innovativet e c h n o l o g i e s ,EDB has set upBio*1 Capital tomanage a $600m i l l i o nBioMedical Sci-ences Invest-ment Fund(BMSIF) andother BiomedicalSciences funds.Bio*1 Capitalplays a key rolein investing inselective compa-nies, commer-cializing indig-enous technolo-

by Ms. Chu Swee Yeok, Director of the Biomedical SciencesGroup; Executive Director of the Biomedical SciencesInvestment Fund, Singapore Economic Development Board (EDB)

A Thriving Biomedical Sciences Industry

Singapore - Best for Business• Best Asian City for Business - EU

2002• Tops in Physical Infrastructure -

PERC 2002• 1st in Asia in Entrepreneurial

Framework - EIU Report• 2nd in Economic Freedom - Washing-

ton based Cato Institute 2002• No. 1 in Labor Force Evaluation

Measure - BERI Report 2001• 3rd Most IT Savvy Nation - World

Economic Forum 2003

gies from local universities, or form-ing strategic joint ventures entitiesin Singapore. To date, it has investedits funds in more than 80 companiesin Singapore as well as overseas.

In addition, other programs have beeninitiated to promote and commercial-ize indigenous technologies in theBiomedical Sciences. These includethe EDB SEEDS (Startup EnterprisEDevelopment Scheme) program whichprovides matching equity funds of upto $160,000 for start-ups in the seedstage of enterprise formation, andthe Biomedical Sciences Innovate ‘NCreate Scheme (BMS INC) underBio*1 Capital, which provides start-up funding specifically to viable busi-ness ideas in the Biomedical Sciences. Under the BMSINC scheme, qualifying companies are eligible to re-ceive up to $1.1 million of seed capital.

At present, Singapore is home to some 30 bio-technology companies, including internationalcompanies like ViaCell and Proligo, as well asa growing number of local start-ups such as S*BIO,CordLife, ES Cell International, and MerLion Phar-maceuticals. This growing pool of both local and inter-national biotechnology companies involved in drugdiscovery and development clearly reflects Singapore’sattractiveness and growth potential as an excellentbreeding ground for new research discoveries to takeoff.

Comprehensive infrastructural support is aggressivelybeing put in place for the BMS sector as well. Expan-sion of prepared land for manufacturing in the TuasBiomedical Park, and a fully integrated R&D complexat Biopolis, which can house more than 2,000 scien-tists promotes physical clustering. This allows foreconomies of scale and significant savings throughshared services and collaboration.

Sectors Employment

2001 2002P % Growth

BMS Total 6,477 7,177 10.8%

Pharmaceuticals 2,375 3,123 31.5%

Medical Technology 4,102 4,054 (1.2%)

2002 Pharmaceuticals Employment (Source: SingaporeEconomic Development Board).[P = Projected]

Manufacturing Value Added ($ million)

2001 2002 % Growth

Biomedical Sciences 2,131 3,746 76

Pharmaceuticals 1,613 3,157 96

Medical Technology 517 589 14

*All figures are based on EDB's preliminary estimates.

2002 Industry Performance: Value-Added (Source:Singapore Economic Development Board).

2002 Industry Performance: Investment Commitments.

Singapore - Valued Partnerfor the Long Term

Within a relatively short span of time, Singaporehas demonstrated significant progress and ison track to achieving its target manufacturingoutput of $6.7 billion by 2005. The governmentis committed to the Biomedical Sciences sector,and will continue to actively provide a soundinvestment climate, a pro-business and vibrantresearch environment for leading biomedicalcompanies and talents to set up base inSingapore.

For information on the ISPESingapore Affiliate visitwww.ispe.org/singapore

Electronic Validation


An Information Based Approach toValidationby Hosam Aleem, Stuart Lord, Tim McCarthy,Paul Sharratt, and Yuyang Zhao

This articlepresents anElectronicValidation(eValid) projectinitiated by aresearch groupat the Universityof ManchesterInstitute ofScience andTechnology(UMIST). Introduction

Validation is a key component of GoodManufacturing Practice (GMP) in thepharmaceutical industry. In additionto being a regulatory requirement, it

also makes good business and technical sense.While the importance of the concept of valida-tion is evident, the practice is problematic.This has prompted a research group at theUniversity of Manchester Institute of Scienceand Technology (UMIST) in the UK to investi-gate this problem. This is taking place throughthe Electronic Validation (eValid) projectfunded by the Engineering and Physical Sci-ences Research Council (EPSRC) in the UKunder the Innovative Manufacturing Initia-tive (IMI).

The objective is to approach validation in astructured and formal way in order to analyzeit and propose solutions aimed at improving itspractice and better integrating it into the over-all business process. It is also intended to quan-tify the benefits gained from applying this meth-odology.

The eValid project is proposing a novel meth-odology that relies on emerging IT standards.The idea is to make the most use of the informa-tion already available early on in the projectlifecycle. This information is increasingly inelectronic form, employing such standards willfacilitate performing a large percentage of thevalidation tasks electronically, or eValid. Workis currently under way, and a demonstratorimplementation is expected to be ready by mid-

Figure 1. A samplebusiness informationmodel of validation.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® July/August 2003, Vol. 23 No. 4



Figure 2. Document content map.

2003. The department of Civil and Construction Engineeringand the department of Chemical Engineering at UMIST areboth involved in this research. The former has a ProjectManagement division, and several staff members with stronginterest in IT and Computer Aided Engineering (CAE), whilethe latter is interested in manufacturing processes thatinclude pharmaceuticals and biotechnology. Industrial col-laborators include major pharmaceutical manufacturers,contractors to the pharmaceutical industry, and softwaredevelopers and suppliers.

The purpose of this article is to inform the practicingpharmaceutical community of the project’s goals and to invitefeedback and comments.

Problems with ValidationIn the years since its introduction, several developmentshave occurred in the practice of validation, but also severalproblems became evident. The prominent ones are summa-rized below.

• The scope of validation is still not clear. The question “Howmuch validation is enough?” doesn’t have a definite objec-tive answer.

• There is significant duplication of effort throughout thelife cycle of a project. Many tasks, tests, and inspectionscarried out during installation, commissioning, and start-up/handover are repeated again in qualification.

• A large volume of paperwork is generated by the valida-tion activity. Validation has become effectively a paperchase.

The approach taken by eValid to address the problems withvalidation focuses on item three above. It proposes movingfrom a document-based paradigm to an information-basedone. It further claims that this step will significantly reducethe duplication of tasks involved in validation. In addition, itis believed that this new paradigm, coupled with other tech-niques to be mentioned later, will lead to a more objectivedecision on the problem of scope. It is interesting to note thatthe problem of documentation is to a great extent caused bythe other two problems. Indeed, the lack of consensus on thescope and requirements of validation leads to an attitude of“better be safe than sorry,” causing more - possibly unneces-

sary - validation work to be done. Similarly, the duplicationof tasks generates even more paperwork. Yet, solving theproblem of documentation will contribute to solving the othertwo.

Below we give highlights of the eValid approach, anddiscuss how it relates to each of the above problems. In whatfollows, validation is discussed as it relates to engineeringactivities, i.e., qualification of facilities, equipment, and au-tomation rather than in cleaning validation or analyticalmethod validation.

Validation: A Paper ChaseThe approach taken by eValid is based on the fact that a largepercentage of the documents - or more precisely information– needed in validation is generated during the earlier phasesof the project. These include the design, construction, andcommissioning phases in addition to product and processdevelopment activities where the product characteristics andprocess conditions are specified. Example documents (infor-mation) include requirements specifications, design and con-struction drawings, material and personnel certificates (e.g.welding and welders), data sheets, operating procedures,manuals, test and inspection reports, etc. Thus, there is alarge pool of information and documents from which most ofthe validation activities will choose, examine, and verify, andwhich is already available, albeit in a non-homogeneous form- Figure 1.

Traditionally, this vast amount of documents has beenmanaged and controlled through some document manage-ment system. To keep matters under control, strict changecontrol procedures are implemented that encompass detailedprovisions for initiating, evaluating, reviewing, and approv-ing the proposed changes. Nevertheless, a problem of contentconsistency is bound to appear at some point in time, becausesome of the document content is likely to appear in otherdocuments which may have been created in other organiza-tions at different points in time - Figure 2. When there aremany such documents, it becomes exceedingly difficult tomaintain an accurate map of this shared content, let alonemaintain its consistency across changes, even in the presenceof a strict change control process. The document changecontrol process places greater emphasis on maintaining theevolution of a given document under control, rather than howchanges in this document relate to and indeed affect otherdocuments, i.e., traceability between documents.

It becomes clear that an alternative view of documenta-tion is needed. This approach will totally eliminate docu-ments and take an information based approach. This mightappear surprising given that documents have long beenregarded as the repositories for information. Indeed, in thepharmaceutical industry if something is not recorded, then ithasn’t been done. However, no claim is made to depart fromthe practice of recording, it is just the concept is now vieweddifferently where the information based approach replacesthe master role of documents with electronic records. Docu-ments will still be available, but are just a snapshot of aparticular collection of information at a particular point intime. What is needed is a more abstract look at the FDA’s



definition of validation, repeated below from its guidelinedocument on process validation.1

“Process validation is establishing documented evidencewhich provides a high degree of assurance that a specificprocess will consistently produce a product meeting itspre-determined specifications and quality characteris-tics.”

The issue here is that “documented evidence” is frequentlyinterpreted as documentary evidence, and thus most of thelegacy documented evidence is taken to be paper documents.However, a document is a set of information. The documentitself will have a format or presentation structure, but thecontent of the document (what the document is about) mayvary from highly structured, like a data sheet which is a setof items (objects or classes) in a table for instance, to unstruc-tured such as free text or an image. Thus, the problem ofcontent consistency mentioned above becomes even moreevident. Note that merely moving from a paper based docu-ment paradigm to an electronic one will not solve the prob-lems inherent in document content management if this moveis not coupled with a change in philosophy. Granted, theelectronic domain will provide more document managementfunctionality, but it does not necessarily solve the informa-tion consistency problem. However, it is a step in the direc-tion of solving this problem, when coupled with means toembed the meaning of the content within the electronicdocument itself. Indeed, an electronic document may be moreor less ‘intelligent’ depending on whether the meaning of thecontent is explicit in its structure or not. An example of a‘dumb’ electronic document is a fax, which is just a bitmap,i.e., a collection of dots on paper, which doesn’t say muchabout the content unless interpreted by a human or anOptical Character Recognition (OCR) system.

In order to solve the document content consistency prob-lem, one should manage at the individual information itemlevel. This requires an identified owner and change manage-ment procedure for individual information items and defini-tion of their meanings. Thus, when each information item iscontrolled, it doesn’t matter in how many documents it willappear as it will always be the same version. The next issuethen would be how to structure and represent this informa-tion such that it can be understood by the different systemsin which it will be needed.

In most modern facility construction activities, an increas-ing percentage of the calculations, drawings, and documentscreated in the design, construction, and commissioning phasesof the project are in electronic form. However, these electronicforms are mostly different and possibly incompatible. Indeed,the software used by the design organization for example,may be different from that used by the construction organiza-tion, consultant, supplier, or regulatory authority, even ifsome of these software products perform the same functions.In addition, many of these electronic documents will eventu-ally have to be transferred to the owner/operator who may yethave a completely different system - Figure 3. Many of the

organizations involved may not necessarily be willing toswitch to a software product used by one of the others, as eachmay have many other customers. Obviously, a supplier can-not change his work system for every customer, especially inthe case of minor jobs. The exception would be when there’sa strategic partnership between the supplier and the cus-tomer, or in the case of suppliers who have multiple systems,and those would probably be larger organizations.

In order to be able to achieve seamless transfer betweendifferent heterogeneous IT systems, two issues need to beaddressed: the meaning of the information and its format.These will have to be independent of any software applica-tion, thus enabling sharing with even yet unknown users.Indeed, this is the case at the early stages of a project whenno bidding has been assigned, and thus, specific users of theinformation are yet unknown. One way to achieve this is toutilize emerging standards that define the meaning of infor-mation explicitly in its structure. One such standard that hasbeen successfully employed in the oil and gas industry is ISO15926 with its different parts, titled “Integration of Life CycleData for Oil and Gas Production Facilities.” In particular,Part 2 deals with the so called “Data Models,” and Part 4 is a“Reference Data Library.”

As for format, the machine readable language in which theinformation is expressed has to be defined. This should be awidely accepted international standard that meets the tech-nical requirements. There are several to choose from, andinevitably the choice is not unanimous. The capability to mapfrom one language to another is required. One obvious choiceis Extensible Markup Language (XML) in addition to meet-ing the above requirements, it has a number of other advan-tages. XML extends HTML capabilities to describe the struc-ture of document content as well as its format. This capabilityallows documents not only to be displayed on the Web (as withHTML), but also to share its content with other documentsand to communicate with other software applications anddatabases. Other advantages include its widespread andgrowing use, its applicability to a Web-based environment,and its inherent power and flexibility yet relative simplicityfor defining semantics. The development of the XML workdone in eValid again follows standards, in this case, those setby the World Wide Web Consortium (W3C).2

Figure 3. Information sharing across systems with differentfunctionality.



the overall context of the business process and how to quan-tify it? These issues will be considered in more detail later. Atthis point, we will consider how much is enough and how todetermine it.

Resuming our earlier discussion of documents and informa-tion, one can take yet another abstract look at the FDA definitionof validation and replace the phrase “documented evidence” by“verifiable evidence.” In the context of the information basedelectronic paradigm proposed earlier, and in view of the recentregulatory requirements concerning electronic records and elec-tronic signatures as stated in 21 CFR Part 113 and elaboratedupon in the related recent guidelines,4 it is evident that elec-tronic records are secure, authorized, and verifiable althoughthey are not necessarily the equivalent of documents. Thus,perhaps the original definition may be expanded to accommo-date different approaches, rather than be restricted to docu-ments. After all, the core of the activity is to verify or prove theconsistency of the performance of the process.

Two other issues need to be resolved concerning verifica-tion. First, what needs to be verified? And second, what levelof verification is necessary to provide “a high degree ofassurance?” i.e., an issue of breadth and an issue of depth.

To determine the scope of validation, and thus to answerthe question of how much is enough from a breadth perspec-tive, eValid proposes basing the answer on a quantification ofthe risks associated. The idea of relying on risk assessment ishardly new, and is recommended by ISPE’s GAMP 4 in thecontext of computer validation.5 However, little seems tohave been done in terms of quantifying these risks, even inGAMP 4 where the approach is more qualitative (high,

Figure 4. Risks relevant to the business and the regulatory authorities.

It should be noted that using such standards and indeedenabling the sensible communication of different IT systemsis not an aim for its own sake, but a means to achieving theimprovement to the validation practice aspired for. Thus, thereader should not lose sight of the ultimate goal in the detailsof the implementation. In fact, most of these implementationissues will be hidden from the end-user in a real life situationemploying this methodology. The end-user here being theuser of this information whether for validation or any otheractivity within the drug manufacturing business process.

We have highlighted above the approach from a practicalstandpoint and how it relates to the end user. However, thereare theoretical issues that aren’t covered here such as theformal modelling and analysis of the validation activity andits information requirements within the overall businessprocess. This was the starting point and was performed usingprocess modelling and reengineering tools.

Validation, How Much is Enough?This question has been debated in industry and the literaturefor many years, and the answer is highly subjective. There isno consensus on this issue, not even among regulatory inspec-tors. Indeed, it is possible to have two firms with differentextents to which they have validated their processes, yet theyboth get regulatory approval, implying that one must havedone more than is necessary. This leads to a fundamentalquestion concerning the consequences or the costs of over-validation and under-validation, even if the company passesinspection. An even more fundamental question poses itselfnamely, what is the value added by the validation activity in



medium, and low). It is not claimed that this is an easy task,and will probably eventually have to rely on subjectivesources such as the estimates of experienced individuals.Nevertheless, quantification adds a level of objectivity that ishoped to increase as the practice matures. It should be notedat this point that the ISPE Baseline® Guide on Commission-ing and Qualification6 addresses the issue of scope, by callingfor impact assessment of the different systems at the onset ofthe project in order to determine what to be qualified. It isthus a notable contribution toward solving this problemalthough the practice still needs to be refined.

The context of risk is another area where the informationbased approach proposed by eValid adds yet more value.While GMP is constrained to risk to the patient from faultyproducts, risks to the environment, employees, and the gen-eral public are excluded. Such risks are still important to thebusiness as well as to other regulatory authorities such as (inthe UK) the Environment Agency and the Health and SafetyExecutive. Risk assessment will rely on information alreadyproduced earlier in the project, and which would be readilyavailable, accessible, secure, and verifiable through the abovementioned approach. Indeed such an approach deals with allthe information of the project/process/ facility and is notlimited to those for validation. Thus, it helps integrate thewhole business process, and facilitate fulfilling all regulatoryrequirements based on such information and not only thosefor GMP - Figure 4.

As for the issue of the level of verification, in addition tobeing related to risk, it is also related to the nature of what tobe verified. This may be a facility, equipment, material,procedure, or personnel. Those terms are used here in ageneric sense and some or all of them may constitute aprocess whether technical or business. In an effort to be morestructured in determining the required level of validationand make it more objective, eValid is proposing a classifica-tion of the different levels of verification. For instance, for aphysical item (material, equipment, facility), one possibleapproach may include the following levels, where the lowernumbers reflect lower levels of verification:

1. Level 1: an item that is a commodity, and thus wouldrequire almost no verification.

2. Level 2: items for which the supplier has to demonstratethe implementation of a quality system such as ISO 9000certification, or by supplying quality records or SPC (sta-tistical process control) charts.

3. Level 3: items for which the supplier would have to supplya certificate, such as a calibration certificate or a legallybinding statement of conformance, this level also mayinvolve supplier audit by the customer.

4. Level 4: items for which an independent third partyaccreditation of the supplier or test of the item is required.This third party may be a regulatory body or a recognizednational or international professional body.

Other levels are conceivable. A similar argument can bemade for personnel, in which case requirements may includean apprenticeship and/or certain years of experience, anacademic degree, or certification by some professional orgovernmental testing body. The point is that different levelsof verification can be associated with the different levels ofperformance required from the object/person/process. An-other dimension of the classification would be related to thecontext where the item will be employed. Thus, the sameclass of item will need to be verified to a higher degree if usedin manufacturing a sterile product than it would be for an oraldosage form for example. The same concept applies to person-nel, even if they have the same set of technical skills. Thisclearly relates back to the issue of risk.

Eliminate DuplicationAs mentioned earlier, many of the activities performed inqualification are a repetition of inspections and tests carriedout in the earlier phases of the project. In particular, thoseactivities include construction, commissioning, Factory Ac-ceptance Testing (FAT), and Site Acceptance Testing (SAT).This has been the situation traditionally, more recentlyhowever, validation is being taken into consideration early onin project planning and is being integrated with the rest of itsactivities. One such approach is the current trend towardapplying Good Engineering Practice (GEP) in all engineeringwork, as recommended by the ISPE Baseline® Guide onCommissioning and Qualification, and indeed the whole ofthe baseline series and other similar good practice guides.

Such good practices producing reliable information, whencoupled with the electronic information handling paradigmproposed by eValid making this information reliably acces-sible and verifiable, can lead to the elimination of much of theduplication currently taking place in qualification activities- Figure 5. Indeed, we propose doing away with all thequalification work that duplicates tasks that have beenperformed in earlier phases of the project, having maintainedthat it was performed according to good practices, and theinformation handled appropriately. In such a case, most of IQand OQ and even some of PQ would become effectively anaudit of these activities performed earlier and not a repeti-tion of them. Obviously, some other qualification and valida-tion activities will still have to be done, especially thoseconcerning the actual performance of the process and testingit under the different realistically conceivable operatingconditions.

Eliminating this duplication raises an important issuethat needs to be addressed, and that is whether the repetitionof testing is necessary for validation. This is meant here on amore fundamental level, i.e., is it a requirement in order forthe practice to be called validation to begin with, irrespectiveof whether it is a regulatory requirement or not. There aretwo possible viewpoints in regard to this issue. The first mayargue that repetition (by a different source) provides inde-pendent verification of the original task, thus providing a“high degree of assurance.” Such a viewpoint can argue thatindependent repetition in validation is almost required by



Figure 5. Duplication of tasks vs. information sharing.

definition. However, the other viewpoint can argue that if thetask is performed correctly the first time, again with a “highdegree of assurance,” then repetition will not add any value interms of this degree of assurance. Indeed, what matters isthat the procedure is performed correctly. If it is performedincorrectly, then no matter how many times it is repeated,that won’t make it valid.

To clarify the latter viewpoint, take as an example acompany procedure which states that an instrument is to becalibrated in a certain range. If the actual operating range ofthe process is not included in this calibration range, then nomatter how many times this calibration is performed accord-ing to this procedure, it will be of limited value from acompliance stand point. On the other hand, if the procedurespecifies the correct calibration range, and the actual calibra-tion is done accordingly, then even if it’s done just once(within the calibration period and without due cause forrecalibration), then it should be in compliance. Furthermore,when a company calibrates its master instrument that it usesas a reference for the rest of the in-house instruments of thesame type and range, it typically sends it to be calibrated ina nationally or internationally traceable calibration lab.After receiving the calibrated instrument, it doesn’t send it

off to another traceable lab to have it recalibrated, why is thisso? One can argue that it is because the “degree of assurance”is inherent in the procedure, rather than how many times itis performed. This clearly ties back to the issue of the levelsof verification mentioned above.

It should be emphasized that these concepts apply to welldefined and well understood tasks that can be unambigu-ously analyzed and tested. Such tasks are fairly simpleconceptually, and are typically parts of IQ and OQ, wheretesting involves checking connections or testing switches,instruments, and controls, etc. However, other tasks haveinherent variability and are not always completely under-stood, and may often involve an interplay of several factors.Such tasks are typically found in PQ, validating sterilefacilities, and similar issues. In such cases, one can’t ignorethis complexity and several runs are required to understandthe interactions let alone verify a property. For statisticalsignificance, repetition will probably be way beyond even thefamous three consecutive runs. Thus, the duplication wesuggest eliminating is that related to qualification activitiesand not the actual validation of the process which typicallytakes place after qualification.

Having mentioned all that, we do acknowledge that the



issue is open to debate and interpretation, and perhaps evento a sense of security (thus subjectivity) that differs from oneindividual and company to another.

The ChallengeIn summary, we believe the proposed approach will simplifyand automate validation documentation, improve the valida-tion process, and integrate it more effectively with the rest ofthe business process. This, in addition to the better definitionof the scope of validation and the elimination of duplicationof tasks, will reduce the time spent in validation and hencetime to market. It is also expected to reduce the costs associ-ated with validation both indirectly by reducing the time andmanpower spent on it, and directly by improving the qualityof the practice. At this point, one would be tempted to askwhether there are any problems associated with this ap-proach. Naturally, there are several challenges involved.

A technical challenge to show that the information basedapproach described above, based on emerging ISO and XMLstandards, will actually improve the quality of information.And, that it can be implemented in an electronic environmentthat is both secure and usable. Also, like any other ITapproach, it has to be validated. These are the areas thateValid is working on.

A cultural challenge also exists as is common with manymajor shifts in concept. Having solved the technical chal-lenge, it has to be demonstrated that the migration to the newenvironment will not increase the risk to the patient, thebusiness, or the regulatory process. Supporting evidencefrom similar (in regulatory terms) industries would be help-ful. A more pronounced cultural challenge; however, is on thepart of the pharmaceutical community - including suppliersand regulatory authorities - to adopt such a shift in paradigm.

Finally, there is an economic challenge to prove thatapplying such a methodology will actually lead to economicbenefit. This is discussed in more detail below.

The Cost and Value of ValidationTo investigate the value added by validation to the overallbusiness process, we need to look at a more fundamentalissue, and that is why do we perform validation. The obviousanswer is that it is a regulatory requirement, but other thanthat there are several business incentives.1. Due to limited sensitivity, some end product testing may

not detect lower levels of non-conformance.

2. Some end product testing is destructive (e.g. sterilitytesting), thus should be reduced to a minimum but thiswould undermine its value.

3. Testing based on sampling can never provide a completedegree of assurance. This is the stance taken by the USPregarding sterility testing.

4. Validation is conceptually a QA function, and as such,provides the many benefits associated with QA such aslower reject, less waste, less rework, and less recall.

5. It also provides better understanding of the process, andhence, can serve as a basis for process optimization.

The above benefits, while making good engineering andbusiness sense, are associated with most QA programs asapplied in any other industry. It is clear that it is theregulatory requirement that makes validation so compelling.Indeed, if validation is not performed and the company failsan inspection, the consequences can be grave, even cata-strophic. This indicates the importance of the cost of valida-tion, more precisely the cost of not doing it rather than thecost of doing it.

One of the goals of eValid was to quantify the economicgains accrued by implementing its methodology. This hasturned out to be more difficult than expected, as the costingof validation to begin with is not so straightforward tocalculate. This may be due to the wide variability in the scopeof validation between firms as mentioned earlier. In the casewhere validation is contracted out, the cost of validation forthe owner is the amount paid to the contractor; however, stillthe costing performed by the contractor needs to be modelledand understood. As for the case where validation is done inhouse, the costing is even more difficult since some of thetasks may be performed as a part time activity by someemployee, or as part of some other task. In addition, there arethe costs associated with the delay in production, the mate-rial used in the validation tests, etc. The eValid team hasdecided to look more deeply into this issue, and has submitteda proposal in collaboration with the Manchester School ofManagement for a project to develop a cost model for valida-tion. It should be noted that previous work has been pub-lished in Pharmaceutical Engineering for a cost model fornon-conformance.7

The Time is RightThe essence of the eValid approach is to treat validation as aninformation management and information quality problem,thus allowing the use of technologies that are being appliedto such problems in other industry sectors. If the methodologycan reduce the volume of paperwork, many of the duplicatetasks will be eliminated. It is also likely that the analysiswork of eValid will result in classification structures whichhave the potential to be associated with risk profiles. This willreduce the areas of personal interpretation, and increaseareas of consensus on how much is enough. We believe thatthe approach developed and proposed by eValid is bothrelevant and timely. First, because there is consensus in theindustry that there are problems in the way validation iscurrently performed and there is potential for improvement.The publication of such guides as the ISPE Commissioningand Qualification Guide and GAMP 4 indicates the presenceof the problems that those guides address. Second, the utili-zation of an information based approach is inline with therecent regulatory trend toward encouraging electronic sub-mission, and enforcing requirements concerning electronicrecords and signatures, 21 CFR Part 11. Third, there is aregulatory trend toward utilizing the concepts of risk man-



agement and in general putting more structure, and thus,predictability in the inspection process. Last but not least, isthe economic factor. The recent or expected expiration ofpatents of many blockbuster drugs, and the lack of develop-ment of financially equivalent replacements have put eco-nomic pressure on pharmaceutical companies. Thus, an ap-proach that reduces time to market would be welcome. Inaddition, companies may consider cutting costs in areaswhere they traditionally might not have considered, such asvalidation, manufacturing, or other activities downstreamR&D. We believe that the approach proposed will meet theneeds and trends mentioned above, and is thus very timely.

References1. FDA Guideline on General Principles of Process Valida-

tion, 1987.2. www.w3.org3. 21 CFR Part 11 Electronic Records; Electronic Signatures.4. ISPE - PDA, Good Practice and Compliance for Electronic

Records and Signatures - Part 2: 5. Complying with 21CFR Part 11, 2001. Part 1: Good Electronics RecordsManagement (GERM), 2002.

5. ISPE, GAMP Guide for Validation of Automated Systems,2001.

6. ISPE Baseline® Pharmaceutical Engineering Guide: Vol-ume 5 Commissioning and Qualification.

7. Dwyer, T., Keresty, G., and Sherry, B., “The Cost of Non-Conformance: The Linkage Between Quality Performanceand Business Results,” Pharmaceutical Engineering,Vol. 20, No. 5, 2000, pp. 8-18.

About the AuthorsHosam Aleem is a Research Associate in theManchester Centre for Civil and Construc-tion Engineering at UMIST. He has an MSfrom the University of Missouri – Rolla andan engineering degree from the University ofSouthern California, both in electrical engi-neering. He is currently studying for an MScin Biotechnology at UMIST. His industrial

experience includes several years as an instrumentation andprocess control engineer in the oil and gas industry, and as avalidation and calibration engineer in the pharmaceuticalindustry.

Manchester Centre for Civil and Construction Engineer-ing, UMIST, PO Box 88, Manchester M60 1QD, UnitedKingdom, email: [email protected], tel: +44-161-200-8966.

Stuart Lord is a Mechanical Engineer bytrade, and has spent his career in the processindustries. He has spent many years in IT,where he has had responsibilities for devel-oping and supporting CAE and business sys-tems as well as managing user support andIT infrastructure. Most recently, he has man-aged collaborative projects, particularly those

associated with the development of standards that define the

meaning of information. For the last few years, he has beenresearching information management in validation, manag-ing a research project at UMIST, and running his consultancybusiness.

Servblock Ltd., Nevada Tempest Rd., Alderley Edge,Cheshire SK9 7BU, United Kingdom, email:[email protected], tel: +44-161-200-8966.

Tim McCarthy is a Senior Lecturer atUMIST, which he joined in 1985. Prior tothis, he worked as a Research Engineer inthe Hydraulics and Maritime Research Labo-ratory at the University College Cork wherehe earned his PhD. He also has spent timeworking with Ove Arup and Partners in theirCork and Paris offices. In 1994/95, McCarthy

worked at Stanford University’s Blume Center for Earth-quake Engineering and Center for Integrated Facilities En-gineering. He holds degrees from UCC, Cranfield University,and is a member of the Institution of Engineers of Ireland.His current research interests include knowledge manage-ment and information management in regulated environ-ments.

Manchester Centre for Civil and Construction Engineer-ing, UMIST, PO Box 88, Manchester M60 1QD, UnitedKingdom, email: [email protected], tel: +44-161-200-4609.

Paul Sharratt joined the department ofChemical Engineering in 1991. Prior to that,he was a process engineer at ICI (now part ofSyngenta). He has been awarded a RoyalAcademy of Engineering /EPSRC chair inInnovative Manufacturing for five years start-ing in 2001. His research interests are inprocess design, environmental management,

and reaction engineering. He has primarily focused on fine/effect chemicals processing and manufacturing strategies(the BRITEST project) and low tonnage chemical processes.

Department of Chemical Engineering, UMIST, PO Box 88,Manchester M60 1QD, United Kingdom, email:[email protected], tel: +44-161-200-4367.

Yuyang Zhao is currently a Research Asso-ciate at UMIST working on the eValid Project.He holds a PhD in chemical engineering fromDalian University of Technology, China. Hisexperience is in the area of operation andmanagement computer modelling for oil andgas, chemical, and the pharmaceutical in-dustry.

Manchester Centre for Civil and Construction Engi-neering, UMIST, PO Box 88, Manchester M60 1QD, UnitedKingdom, email: [email protected], tel: +44-161-200-8966.

Document Retention Policy


From Never to Forever: DocumentRetention Policyby Mark Kropp, MD

This articlesupports thewriting of adocumentretention policyand givesgeneral timeretentionperiods fordifferent kindsof documentsincludingelectroniccopies. Itoutlinesplanning forretention andcan be usedwhenprogrammingcurrentcompliancesoftwarepackages.

Introduction

Document retention is a significant tech-nical challenge for the pharmaceuti-cal industry as well as all businesses.Many of the Knowledge Management

Engineering Software Packages sold as com-pliance tools require decisions regarding docu-ment retention. Because of the importance ofthe issue to pharmaceuticals and customers, aswell as the potential for disputes or litigation,document retention is a “business necessity.” Itinvolves both company internal systems aswell as projects that the company has under-taken on behalf of clients. However, the pur-pose is not to retain every piece of paper orelectronic information ever created during thecourse of a project, but to retain the documen-tation which allow the company to respond tocustomer, client, and government inquiries, orto show what the company did on a particularproject and that the company complied with itsobligations.

Competitive pressures, government regula-tions like Health/Insurance Portability andAccountability Act (HIPPA), Gramm-Leach-Bliley, 21 Code of Federal Regulations (CFR)Part 11, and recent media coverage of sensitivecorporate e-mail being exposed, are all drivingorganizations to focus on secure electronic docu-mentation. One of the many lessons learnedfrom the “Anderson trial,” with its focus ondocument shredding and the prevalence of suchelectronic evidence as emails, is not just thepotentially incriminating nature of electronic

archives, but the liability of inadequate en-forcement of a document retention policy. It’sone thing to have a policy; it’s another to imple-ment and audit it.

On July 30, 2002, the President signed intolaw the Sarbanes-Oxley Act of 2002. Enacted inresponse to the recently exposed corporate andaccounting wrongdoing, the “Act” contains someof the most significant changes to the federalsecurities laws since their enactment duringthe Depression. The “Act” creates a new federalaccounting oversight body; revamps auditorindependence rules; enacts new corporate re-sponsibility and governance measures; en-hances disclosures by public companies; regu-lates potential conflicts of interest by securitiesanalysts; strengthens the powers and resourcesof the Securities and Exchange Commission,and imposes new penalties for securities fraudand related wrongful conduct.

Information ProtectionAn effective information protection programcannot be solely defined in terms of trust. Rather,it must be based upon the same prudent busi-ness practices that applied to earlier manualsystems and statement/publication of policy,careful definition of individual responsibili-ties, separation of controls, maintenance ofaudit trails, protection of vital records, andaccess to limited information, based on “need toknow.” These are all controls, and are exactlywhat auditors look for.

Table A. Documentretention decision tree.

Is there a legal requirement to retain the document? Yes No

Is there a use for the document after its intended use? Yes No

Is there a consequence for not being able to locate the document? Yes No

Can the document be reproduced? Yes No

Can the document be retained? Yes No

Is the document important for pending or threatened litigation? Yes No





Some of the positives that can begin to result from publish-ing document retention policies/guidelines are:

• greatly improved document retention focus• fewer litigation incidents• fewer audit concerns/comments• greatly improved business focus• enhancement of the professional perception of the busi-

ness• fostering of a team oriented environment• enhancement of employee morale• helpful in attracting and retaining the best people

Some objectives are:

• control who can see information and whether they canprint, copy, or select text

• prevent information from being forwarded• recall or expire information, even after it’s accessed• track what recipients do with your information after they

download it

Responsibilities and PolicyThe Sarbanes-Oxley Act of 2002 requires enforced recordsmanagement programs at US companies and shifts personalaccountability to executives:

Title VIII: Corporate and Criminal FraudAccountability Act of 2002.It is a felony to “knowingly” destroy or create documentsto “impede, obstruct, or influence” any existing or con-templated federal investigation. Auditors are required tomaintain “all audit or review work papers” for five years.The statute of limitations on securities fraud claims isextended to the earlier of five years from the fraud, or twoyears after the fraud was discovered, from three yearsand one year, respectively. Employees of issuers andaccounting firms are extended “whistleblower protec-tion” that would prohibit the employer from takingcertain actions against employees who lawfully discloseprivate employer information to, among others, partiesin a judicial proceeding involving a fraud claim. Whistleblowers are also granted a remedy of special damagesand attorney’s fees. A new crime for securities fraud thathas penalties of fines up to 10 years imprisonment.

Companies, in anticipation of potential litigation, shouldhave a document retention policy set up, and they need tooperate within a policy created by their lawyers and to enforceit regularly so that they can be certain they are operatingwithin the realm of the law and protecting themselves frompotential harm.

Not having a policy, or having one, but not acting on it ona regular basis is a problem, as was illustrated by the“Anderson trial.” If a document retention policy is acted ononly before pending litigation, a company’s actions may nothold up in court.

Preventative maintenance, including the education andtraining of employees on the policy, is essential to ensure thepolicy is enforced. Management and counsel should worktogether to test the effectiveness of the policy by conductingperiodic searches of the data environment to see whether ornot anything of interest turns up. If something is found,counsel and the client discuss the ramifications and developa strategy for dealing with that data or problematic behaviorbefore anything gets to the point of litigation so that the firmis protected and doesn’t incriminate itself by keeping need-less files that it has a right to eliminate.

Destroying incriminating evidence or unethical behaviorisn’t implied here, rather data that isn’t official communica-tion, such as working drafts, and day to day email with nofuture value, certainly, if there is anything that could beperceived as a “smoking gun,” it is better to know about it, tominimize litigation risks. It’s critical that companies knowthe contents and mange their information archives. Doing soforces employees to prioritize, to conserve network storage,and to conduct themselves ethically.

If a policy exists, it needs to be audited. If one says theseare things done and aren’t, how does one know employees arefollowing the policy? Does one want to put InformationTechnology (IT) departments in the difficult position of audit-ing and scrutinizing the integrity of co-workers? One needs toperiodically pull in a third party firm to audit adherence to“communication” policy. Recent events and trends suggestthat as firms become involved with lawsuits, business lead-ers begin to appreciate the value of managing the risk.Insurance rates are going up, and eventually companies arerequired to enforce and audit their document retention poli-cies with third party risk management firms in conjunctionwith attorneys.

Although this is a cost containment and risk managementexpenditure for corporations, businesses need to understandit is a critical one, because the costs of not developing,enforcing, and auditing a document retention policy could bedevastating.

A document retention policy is a set of guidelines that acompany follows to determine how long it should keep records,including email, web pages, quality documents. The policy isimportant for many reasons, including legal requirementsthat apply to some documents. Why and how long is the nextquestion. Occasionally clearing away unused items is neces-sary; however, tossing the wrong paper or deleting a criticaldata file can have dire consequences, especially for pharma-ceuticals.

Document retention policies can range from a few para-graphs to many pages. A good document retention policyanswers the question: “What can I throw away (delete), andwhen?” Some policies contain detailed instructions for wheredocuments will be kept, the type of storage container, and themanner of disposal (such as shredding). Most policies providea list of the types of documents produced and how long thosedocuments should be kept. For example, our Clinical Re-search Organization (CRO) contracts five year retention withsubsequent returning of all documents to the client.



Today, a good pharmaceutical document retention policydeals with electronic files and electronic mail. The objectiveof a document retention policy is to reduce the volume ofpaper in storage, or data on disk, meet legal requirements forrecord keeping, and stem paranoia.

Six Issues for ConsiderationHow long to keep a document, when and how to store thedocument, and how to dispose of the document will depend onthe type of document – Table A. Six issues for considerationare:

1. Is there a legal requirement for keeping the document?Legal requirements include federal, state, and local re-porting concerning various regulated matters, such aswages and hours, health and safety, shipment and han-dling of hazardous materials, quality and engineeringdocuments.

2. After the item is used for its intended purpose, what otherpurpose could it serve? Can it be used to support or opposea position in an investigation or litigation? Can it supporttax deductions? Is it used in application for Food and DrugAdministration (FDA) approval?

3. What is the consequence of not being able to locate thedocument? If the document was destroyed pursuant to arecords-retention program and no threat of litigation waspending at the time, the issue will be how reasonable theprogram was. If the item was mentioned in a lawsuit, thensuddenly destroyed, the presumption will be that thedestruction was accomplished deliberately.

4. Can the item be reliably reproduced elsewhere if needed?Is the information available from the public library, anonline source, a database, or company central file? Forexample, our CRO sends the same memorandum to mul-tiple recipients and each does not keep a copy.

5. Once the possible use of a particular item is determined,the question becomes how long to retain the document.This question is answered by reviewing the statute oflimitations (the time within which a suit must be broughtfor a particular action after the action is discovered) inboth state and federal government regulations.

6. If the document is in any way related to pending orthreatened litigation, it is wise to keep the item until thematter is finally settled or all appeals are exhausted.

The following examples are drawn from the web. The recom-mended retention period exceeds the required statute becausethe limitations period for litigation is longer than the statutoryrecord-keeping requirement.

Email and Web PagesThe length of time email should be retained depends on the

content. After an email is forwarded with the “latest joke” toall, it can probably be deleted (unless it’s one of those kind ofjokes, at which point it will likely be attached to a complaintfor harassment). It is suggested that a hard copy of importantemail be kept in the file to which it pertains. It is a good ideato have backup documentation, especially if a system may fail(crash). Some systems are archived immediately, overnight,on weekends. While this may preserve a snap shot of thesystem, most backups are lost when the same media is usedfor the next backup. In other words, don’t rely on the fact ofa backup to preserve important email. It is a good idea topreserve each iteration of a web page, especially if a disputearises.

Employee Records and Employment/Training Manuals

Employee records should be retained for the length of theemployee’s tenure with the company, plus at least the statuteof limitations period. Many policies require records to be keptfor at least seven years. Payroll records should be stored forthe same period as tax records. Most state laws require thatany action by an employee based on discrimination must firstbe filed with the Equal Employment Opportunity Commis-sion or the state equivalent within 180-300 days of the actgiving rise to the complaint. A formal lawsuit in state orfederal court may be filed after the administrative agency hascompleted its review or within six months of the administra-tive filing. Most states allow these types of suits to be filedwithin two years of the act, giving rise to the complaint. Othercauses of action, such as breach of contract or various tortactions (such as infliction of emotional distress or wrongfuldischarge) have limitation periods varying from one to sixyears. Suits can last for many years, and the documents mustbe preserved throughout the suit. Because employment recordscan contain very sensitive information, they should be storedin a secure area. Certain types of records, such as theImmigration and Naturalization Services I-9 form shouldactually be kept separately from active employee files toavoid claims of national origin discrimination. When thesedocuments are ready for destruction, they should be shreddedto avoid disclosure. A copy of each version of employment andtraining manuals should be kept with the dates that versionwas in use. The reason for, or timing of, a change in themanual may become important in a suit.

Sales Documents IncludingRecords and Presentation Materials

The recommended retention is the length of the sale plus thelimitation period. Many policies require that such records beretained for anywhere from three to seven years. The recordsshould be kept, unless the information can be reproducedelsewhere, for as long as needed to protect the company in theevent of legal action. Correspondence leading up to a sale, aswell as the sales materials that resulted in the sale, mayserve as evidence of promises made to make the sale. Solici-tation letters for products or services not purchased need notbe retained. Like employment manuals, a copy of each ver-



sion of sales document should be kept with the date thatversion was in use. Some companies keep market researchand projections for as long as 20 years, as they serve thepurpose of historical comparison. This points up the need toevaluate the type of document and determine how to treat it.

Tax Related DocumentsThe recommended retention is seven years. If tax audits goback three years from the date the return is filed, and sixyears from that date if fraud is suspected, taxes are filed eachyear after the tax was incurred so keeping the supportingdocuments for seven years should cover audits.

Real Estate RecordsThe recommendation is to check with an attorney in the stateof the holding, but on average, 20 years because the statuteof limitations for real estate is long in most states and oftento prevent squatters from claiming possession of land byreason of having lived on the land for an uninterrupted periodof time this time is needed. This means the actions concerningreal estate can be brought in some cases up to 20 years afterthe action arose.

21 CFR Part 11Subpart B - Electronic Records11.10 Controls for closed systemc. Protection of records to enable their accurate and ready

retrieval throughout the records retention period. (Goes topredicate rules).

Medical DevicesGoes to product liability and government regulation. Thesaying is “one year from forever;” however, the life of theproduct is also a timeline often used.

InventionsThe recommended retention is permanently. An intellectualproperty registration (patent, trademark, copyright) can al-ways be challenged. Documentation relating to the date theinvention was conceived, the trademark first used, or thecopyrighted item first published, can be vital.

SummaryA. Recognize that implementing a document retention policy

has a legitimate purpose. If there is no business reason tokeep a document and no legal obligation to retain it, it canbe destroyed, as a matter of practice, in order to reducestorage costs.

B. Ask an attorney to write the policy to assure the criteriachosen (usually based on the date the document wascreated) are legally correct. The purpose of the policy is notto evade the law or to create a legal problem. The purposeof a legitimate document retention policy is to manage,properly and legally, the mass of documents that a corpo-ration generates.

C. There is no wisdom in “aggressive” or “clever” legal posi-tions when adopting and implementing a document reten-tion policy. Current events demonstrate the price of thefolly and that paid by those who do.

D. Do not destroy documents if a legal “matter” is pending.Clearly, a court proceeding is a “matter.” A grand juryinvestigation is a “matter.”

E. Do not distribute a document retention policy withoutexplanation as how to apply it. Any document retentionpolicy should be accompanied by written guidelines onhow to apply it along with periodic training on its properuse.

F. Do not implement a document retention policy in haste ata time of crisis. This will become the worst time andapproach. Also, worse then a bad document is an illegallydestroyed document.

G. Teach how to properly prepare internal documents. Retainthose data required by law. Finally, do not let the law bemisstated or allow any erroneous jumping to legal conclu-sions.

About the AuthorMark Kropp, MD is Manager of ComputerValidation at In Vitro Technologies, Inc. basedin Baltimore, Maryland. He has more than15 years of validation experience includingworking with Bayer, Alza, and Johnson andJohnson. He can be reached at (410) 455-1242.

In Vitro Technologies, Inc., 1450 S. Roll-ing Rd., Baltimore, MD 21227.

21 CFR Part 11


Adopting a Risk - Based Approach to21 CFR Part 11 Assessmentsby Ken Phoenix and John Andrews

This articledescribes howto adopt a risk-based approachto 21 CFR Part11 (Part 11)assessments forcomputersystems usedfor cGMPrelevantactivities withinthepharmaceuticalindustry.

Background

In March 1997, the FDA issued final regula-tions to 21 CFR Part 11 (Part 11) thatprovided criteria for acceptance by the FDA,under certain circumstances, of electronic

records, electronic signatures, and handwrit-ten signatures executed to electronic records asequivalent to paper records and handwrittensignatures executed on paper. These regula-tions, which apply to all FDA program areas,were intended to permit the widest possibleuse of electronic technology, consistent withthe FDA’s responsibility to protect the publichealth.

The final regulations became effective inAugust 1997, and since that time, they havebeen the subject of ongoing debate within thepharmaceutical industry on interpreting howand when the regulation should be applied.

Industry concerns center around:

• restriction of technological innovation• lack of clarity on how the regulation should

be applied for specific applications• significant cost increases for implementing

computer systems within the cGMP-relevantenvironment

The topics for debate centered on the Part 11requirements for validation, audit trails, recordretention, record copying, and legacy systems,and the situation was made worse by com-ments from the Agency’s staff being misinter-preted as FDA Policy.

There also was significant ‘interpretation’and individual assessments that muddied thewaters further – it was never intended that itshould restrict technological innovation noradd a significant cost burden on the computersystem validation and cGMP compliance ac-tivities.

For the industry, the introduction of Part 11

heralded a significant growth in the Computervalidation work, and some members of staffnow have primary responsibility for Part 11and are industry-recognized experts in the field.Don’t undervalue them, your trained resourcesstill have an important job to do in helping youremain in compliance with Part 11.

The cost of achieving a degree of indepen-dence from external expertise has been signifi-cant for many pharmaceutical companies, andtoday, they have the opportunity to reap thebenefits that Part 11 compliance can bring,including:

• start to realize the vision of ‘going paperless’• added security• knowing who was at the controls during any

stage of production• having the audit trail, electronic records and

signature capabilities required by regula-tion

• reduction of costs by eliminating unneces-sary paperwork

• quicker NDA submissions and faster time tomarket

• higher degree of quality and consistency –reduced human error

• ability to view data in context with auto-matic management and incident report gen-eration

• smoother regulatory inspections

A New Direction: A Risk-BasedApproach and Smart Regulations

Faced with industry concerns and an ever-growing burden of undertaking regulatory in-spections, the FDA announced a highly signifi-cant change of direction in August last year toimprove regulation within pharmaceuticalmanufacturing.

The FDA’s Health and Human Services(HHS) Secretary; Tommy G. Thompson has set



21 CFR Part 11


out to introduce sweeping reforms in the HHS regulationwith the stated goal of introducing smart regulations toimprove access to quality healthcare and services.

Mark B. McClellan, MD, Commissioner of Food and Drugsunderpins this “These initiatives are Part of the Departmentof Health and Human Services’ overall efforts to improve thequality, safety, and cost of medical products.”

The initiative “Pharmaceutical current Good Manufactur-ing Practices (cGMPs) for the 21st Century: A Risk BasedApproach,” is a two-year program which applies to pharmaceu-ticals, including biological human drugs and veterinary drugs.

Mark McClellan’s statement, “We will focus our attentionand resources on the areas of greatest risk with the goal ofencouraging innovation that maximizes public health protec-tion and promotion,” portrays the essence of the change.

Consequently, in a move that surprised much of thepharmaceutical industry, the Federal Register of February 4,2003, announced the withdrawal of the draft guidance forindustry, 21 CFR Part 11; Electronic Records; ElectronicSignatures, Electronic Copies of Electronic Records.

Since this announcement, confusion has been expressedand strenuous efforts by the FDA to clarify their intent hasresulted in yet more confusion.

Most importantly, the FDA has not withdrawn Part 11 andrecords must still be maintained or submitted in accordancewith the underlying predicate rules.1

There are significant implications in the detail, but thebroad aims of the FDA when they withdrew the Part 11Guidance are simple enough to understand:

• to facilitate innovation for modern manufacturing, elec-tronic record keeping, and regulatory submissions, andallow manufacturers to make certain types of changes intheir processes without prior FDA approval

• to exercise enforcement discretion with respect to certainPart 11 requirements while FDA considers whether torevise the Part 11 regulations

• The FDA will not normally take regulatory action toenforce Part 11 with regard to systems that were opera-tional before August 20, 1997 while FDA considers whetherto revise the Part 11 regulations.

• Part 11 will be interpreted narrowly, and subject to FDAclarification, fewer records will be considered subject toPart 11.

The new draft, “Guidance for Industry Part 11, ElectronicRecords; Electronic Signatures - Scope and Application,” is-sued in February this year is currently available for industrycomment. This Guide replaces all other Part 11 guidance andthe FDA’s Part 11 enforcement policy.

This draft guidance attempts to define:

• Narrow Interpretation of Scope - stating that the merelyincidental use of computers would not trigger Part 11

• Definition of Part 11 Records - recommending that, foreach record required to be maintained by the predicaterules, you determine in advance whether you plan to relyon the electronic record or paper record to perform regu-lated activities and document your decision. Businesspractices also may be taken into account to determine ifPart 11 applies.

• Validation - should be based on the predicate rules andensuring the reliability and accuracy of the Part 11 records.The FDA recommend that you base your approach on ajustified and documented risk assessment to determine ifthe system will affect product quality and safety andrecord integrity

• Audit Trail - where the FDA plans to exercise ‘enforcementdiscretion’ regarding the specific Part 11 requirements forcomputer-generated, time-stamped audit trails. Again, adocumented risk assessment is recommended when con-sidering the design of the audit trail.

• Legacy Systems - where the FDA plans to exercise ‘en-forcement discretion’ regarding the specific Part 11 re-quirements with regard to systems that were operationalbefore August 20, 1997. However, all systems must complywith all applicable predicate rule requirements and shouldbe fit for their intended use.

• Copies of Records - that defines the requirements forcreation, copying, and review of the records

• Record Retention - where again ‘enforcement discretion’ isintended. A documented risk assessment also is recom-mended when considering the design of the system forprotecting the records throughout the retention period.

There is much in the new draft guideline that requiresquestioning (and has been), and key terms such as ‘enforce-ment discretion,’ unfortunately, remain undefined.

However, what is significant is the repeated reference to a‘documented risk assessment’ in the recommendations, buthow should this be done?

Example of a Risk-Based Approach to Part11 - Building Management System

The control systems associated with building environmentalmanagement, typically known as Building Management Sys-tems (BMS), have always presented a difficult challenge tothose responsible for validation. This is because cGMP andnon-critical facilities are generally housed in the same build-ing. Therefore, the control systems have generally beenmixed, thus making it very difficult and expensive to vali-date. Segregating the control system between cGMP and non-critical also is very difficult because the air-handling equip-ment and other such equipment may be common to bothfacilities. They are often considered to be too difficult tovalidate, but the regulators are unconvinced. A typical FDA

21 CFR Part 11


Warning Letter illustrates the point:

The alarm system that communicates, records, andcontrols alarms such as air balance and temperaturesfor production, warehouse and testing areas lackedvalidation documentation (FDA Warning Letter, Janu-ary 2001).

A typical Building Management System is shown in Figure 1.The system unifies the environmental and security data

from the manufacturing plant, cleanrooms, gowning rooms,physical points of access, etc., for presentation to the plantmanagers and operators. Electronic records are generatedand stored on a secure server for future regulatory review.

Specific areas of importance from a Part 11 viewpointinclude:

Security - Physical and Logical• Typically, the standard ‘two-token’ username and pass-

word combination are used within a Part 11 compliantenvironment for logging onto the BMS computer systems.

• Physical security of access to the facility may be providedby swipe cards, proximity cards, biometrics, and videosurveillance, or a combination of these.

• The requirements for adequate physical security are speci-fied in 21 CFR Part 11.10 that defines the measures toensure the authenticity and authority level of personnelwith access to restricted areas and workstations.

• The system is configured to generate an alarm if anattempt is made to gain unauthorized access or initiate alockout if there are multiple failed access attempts. Thiscan then be recorded in a report for the inquiry team.

• The movement of personnel and computer log-on/log-offmust be recorded and reproduced using electronic records.Furthermore, access restrictions can be introduced inspecific zones of the plant.

• Within the computer systems log-on security can be tiered,e.g., operator, supervisor, manager, engineer, such that

Continued on page 76.

Figure 1. A typical building management system. (Image courtesy of Andover Controls Europe.)

21 CFR Part 11


individuals have pre-defined access rights to the systemfunctionality.

• Part 11 extends to protecting the environment in whichthe records are stored, e.g., secure server and archiveroom.

Environmental Controls• Controlling the nature and quality of air in a manufactur-

ing facility can be of paramount importance and devia-tions can dramatically adversely affect the end productquality.

• The ‘Environmental Control’ aspect of the BMS system istherefore a cGMP-critical attribute of the overall system.

• The control systems can be very complex, going far beyondthe simple temperature and humidity control provided bysimple HVAC systems.

• The controlled parameters could be temperature, humid-ity, particulate counts, differential pressure, lighting, gaslevels, etc. This extends to the laboratory where additionalequipment may be required to detect toxic gases and fumehood positions.

The BMS-controlled and monitored parameters are recordedelectronically to provide the operational staff with a view ofcurrent conditions and provide evidence of the overall qualityand safety history of the facility.

The data from the BMS system can be analyzed andcorrelated, e.g., a temperature alarm can be traced to anindividual entering part of the facility and holding the dooropen, and the alarm and event reports can be compiled toconvey meaningful information.

For a BMS system, the following topics might be includedin the risk assessment:

• uniqueness of username/password token• access rights for different employee types• means of obtaining physical access - swipe card, biomet-

rics, etc.• audit trail for electronic records• security of access for records• server security• data archiving and retrieval - media type and storage

conditions• user log -on and log -off audit trail• authorization to cancel alarms and warnings• periodic review

The above gives just a few examples; a definitive list wouldneed to be compiled for each application.

Executing the Risk AssessmentThe principles of risk assessments can be summarized asfollows:

1. Identify the potential risks.

2. Assign the inherent severity (worst case impact) andprobability (assessment of the likelihood of the eventhappening) associated with each risk, e.g., high, medium,and low severity; high or low probability.

3. Identify measures that can be taken to reduce the impactof high/medium severity and/or high/medium probabilityrisks. (The objective is to design measures that, ideally,reduce both the probability and severity to ‘low – low’).

4. Assign the expected residual severity and probability toeach identified risk after the corrective measures havebeen introduced.

For example, a data server may be currently located in anopen office - the inherent probability of someone tamperingwith it is high and (as it contains cGMP-critical data) theseverity (impact) of unauthorized access also will be high.This is a high-risk installation for Part 11 compliance.

The control measure would be to re-locate the server in asecure room with restricted access.

The residual probability and severity of the risk shouldthen be ‘low-low’ assuming that the new location has beencorrectly designed and it is used as intended.

While the risk assessment process is simple in principle,the reality of executing it can be more complex.

For example, the desired control measure may not becapable of implementation for practical reasons or cost con-straints - alternative approaches will then have to be devised,e.g., additional procedural controls administered by the com-panies quality assurance functions.

Another situation that arises is the introduction of second-ary risks that are introduced as a consequence of introducinga control measure. Using the simple example above, re-location of the server may have removed it from an areawhere Uninterruptible Power Supply (UPS) support wasprovided and an additional UPS may have to be purchased toremove the secondary risk.

There are a number of standard risk assessment tech-niques available and the Failure Mode Effects Analysis(FMEA) approach, for example, is widely used within theindustry. The following approach has been developed specifi-cally for the data management requirements of 21 CFR Part11 where the probability of the risk arising and the probabil-ity of detecting the error are the risk assessment parameters.

After completing the following three steps, the recordsdeemed high/medium risk from the results of the risk assess-ment should then be further assessed against audit trails andrecord retention requirements of relevant predicate rules.Using this assessment tool looks easy, but will highlight thegaps in normal operational expectations to comply with thenarrowed interpretation of Part 11.

Step 1 - Does the System Impact Part 11?Does the system manage, store, or use GxP electronic records?Y/N

21 CFR Part 11


Consider:Are the records required by predicate rules and main-tained in electronic format? Also, are the records requiredby predicate rules maintained in electronic format andpaper format where the electronic format is relied on toperform regulated activity?Note: Review business practices to ensure the electronicformat of a record is or is not performing a regulatedactivity. Is this document in an SOP?

Does the system impact Predicate Rule requirements? Y/NConsider:Was the system in place before August 20, 1997? If theanswer to the question is Yes, Part 11 may not apply.Review current use against predicate rule requirements.If the answer to the above question is Yes, has there beenany major upgrades made to the system since August 20,1997? If Yes, Part 11 may apply.Note: Are records in electronic format in place of paperformat, if Yes, Part 11 would apply?Or: Is the system used to generate paper print outs ofelectronic records, and those records meet the require-ments of the predicate rules, and persons rely on the paperto perform regulated activities; if Yes, Part 11 would notapply? Is this document in an SOP? Also, what happens tothe electronic record?

Is the system used to approve and/or authorize GxP opera-tions, or to authenticate GxP electronic records by means of anelectronic signature or other electronic mechanism? Y/N

Note: Paper and e-records and signature components canco-exist as long as predicate rule requirements are metand the content and meaning of the records is preserved.

If the answer to the above three questions is No, then Part11 does not apply.

Step 2 - Risk ManagementProduce a process flow diagram identifying all major func-tions, interdependencies, i.e., network connections, othercomputer systems, and peripherals like printers, interfaceswith people including the SOPs.

• What are the major GxP functions and associated perfor-mance requirements of the system? - list all the majorsystems functions and any performance criteria; this in-formation can be derived from the flow diagram and theUser Requirement Specification/Functional Specificationfor the given system.

• From the major functions, what GxP data is produced andhow does it impact predicate rule requirements - thisinformation can be derived from the systems design docu-ments.

• If the system fails to perform a function that impacts onpredicate rule requirements correctly, what are the failureevents? - from the list of major functions, look at thedifferent types of failures that may exist in the operatingenvironment.

• What is the effect on GxP of each failure event? - assess ifthere is an impact GxP for each failure event.

• What is the probability of each failure effect being detected?- categorize into low, medium, or high probability ofdetection in a normal production environment.

Note - above example for illustration purposes only

Table A. Risk Assessment - example using an eCRF application.

Majorfunctions?

Patient history file

What GxP datais produced?

Baseline data recordingpatient 1st visit andhistory information

Study data results ofall subsequent visit andtest results

Visit data lists thenumber and dates of allplanned visits and tests

Failure events -identifythe risks

Incorrect baselinedata recorded

Baseline data lost

Incorrect studydata recorded

Study data lost

Incorrect visitsscheduled

Visit historymissing

What is the effect on GxP of eachfailure event?

Incorrect dose set

Incorrect study result

Study delayed

Patient removed from study

Study results wrong

Study abandoned

Study delayed

Study abandoned

Study results wrong

Patient removed from study

Impact onGxP (Y/N)

Y N

Probability ofDetection?

L M H

21 CFR Part 11


Part 11 Guideline comment

You should provide theinspector with reasonable anduseful access to records duringan inspection

Provide copies in commonformat where records are keptin these formatsOr - using establishedautomated conversion methodsto make copies into a morecommon format

If you sort, trend etc.; copies tothe agency should also have thesame capabilityConsider procedures andtechniques to access records

Table C. Risk Assessment Report.

Section

11.10(b)

Preamble ref.

69, 70

Questions to Consider

11.10 (b) The ability to generate accurate and complete copies ofrecords in both human readable and electronic form suitable forinspection, review, and copying by the agency. Persons should contactthe agency if there are any questions regarding the ability of the agencyto perform such review and copying of the electronic records.

• Can a copy of a single record (in electronic format) be supplied to aninspector? In paper format?

• Can a copy of the entire database (in electronic format) be supplied to aninspector?

• Are procedures in place to describe HOW to accomplish these inspectiontasks?

• Are procedures in place to define what format the electronic records will beprovided?

Probability of Detection

High Medium Low

High Medium Priority High Priority High Priority

Medium Low Priority Medium Priority High Priority

Low Low Priority Low Priority Medium Priority

Table B. Risk prioritization.

Prob

abili

ty o

fit

Happ

enin

g

• Assess the probability of each failure effect happening -categorize into low, medium, or high probability of ithappening.

• What modifications to the design or enhancements to SOPscan be made to reduce GxP risks – Review findings andmodify design to eliminate the high risk/high probabilityof it happening functions. Enhance SOPs to cover lowerpriorities. Use a Part 11 checklist to assess system compli-ance and likely resolution requirements - Table A.

In Table B, the baseline data and the study data are consid-ered to be high priority if there is a medium to high probabil-ity of it happening and a medium to high priority if there wasa low probability of it happening. Therefore, it is important toassess the full compliance status of this system and addressany compliance deficiencies in relation to handling baselinedata and the study data.

Having established the priorities, work can commence ondesigning the necessary corrective measures. When this isdone, the risk assessment can then be re-executed to deter-mine the residual risk - this is an iterative process wheresecondary risks may be identified along the way. When therisk assessment team is satisfied that they can achieve a‘minimum risk’ solution, the Risk Assessment Report can becompiled.

Step 3 - Part 11 AssessmentConduct a 21 CFR Part 11 assessment of the system using astandard checklist, e.g., using the ISPE GAMP Guide, to

assess the likely remediation requirements to meet fullcompliance.

The structure of the Risk Assessment Report should clearlydocument the process you followed and it helps if you includethe Part 11 requirements together with the questions youneed to consider, as shown in Table C.

ConclusionsFollowing the principles described above should help guidethe reader through a logical risk assessment, and hence riskmanagement approach to compliance with 21 CFR Part11 inthe context of the new direction being adopted by the FDA.The proposed changes to Part 11 are still in the draft/consultation phase, but the future direction the FDA wants tofollow is already clear.

A clear, logical, approach to managing Part 11 compliancethat has been correctly documented and followed through willhelp avoid difficult questions during your next inspection.

References1. The underlying requirements set forth in The Federal

Food, Drug, and Cosmetic Act (the Act) Public HealthService Act (the PHS Act) and FDA regulations are re-ferred to as the predicate rules.

2. Regulation: 21 CFR Part 11; Electronic Records; ElectronicSignatures, Electronic Copies of Electronic Records. - FDA.

3. “Pharmaceutical current Good Manufacturing Practices(cGMPs) for the 21st Century: A Risk Based Approach” -FDA.

4. Draft: “Guidance for Industry Part 11, Electronic Records;Electronic Signatures - Scope and Application” - FDA,(issued Feb. 2003 with 60 day review period).

5. White Paper: “Smart Facility Automation Solutions ForRegulatory Compliance” Andover Controls(www.andovercontrols.com).

21 CFR Part 11


6. GAMP 4, GAMP Guide for Validation of Automated Sys-tems, ISPE (Publishers), 2001.

About the AuthorsKen Phoenix, B.Sc., recently joined GxPfrom KMI-PAREXEL where he worked inSweden on CSV projects for a major pharma-ceutical/biotech company. Phoenix is nowthe Computer System Validation Group Man-ager for GxP. With a background in largeprocess control and IT projects, he has beenworking exclusively on computer system vali-

dation projects for the past three years. He was born inBirmingham and has 16 years of experience in the pharma-ceutical/biotech industry gained within Europe.

John Andrews is a Principal Consultant atthe Synapse Partnership Ltd., Manchester,UK, and a member of the GAMP 4 SpecialInterest Group on Process Control. He also saton the Editorial Board for GAMP 4. His re-sponsibilities include providing consultancyon computer systems validation, compliance,and quality assurance activities within the

pharmaceutical, biopharmaceutical, medical device, and otherregulated healthcare industries. Previously, he was managerof IT Consulting Service at KMI, a division of PAREXRELInternational LLC, and held positions as computer systemvalidation manager and supply chain systems project man-ager with GlaxoSmithKline. Responsibilities atGlaxoSmithKline included: all aspects of computer systemsvalidation, from process control through to business and labo-ratory system validation. He managed the teams with respon-sibility for ensuring all computer system validation activitiesundertaken on site and within projects were to an appropriatelevel to comply with the regulator’s requirements. Previousemployment includes 15 years with SmithKline BeechamPharmaceuticals where he held positions as a senior engineer-ing standards engineer, secondary manufacturing electricalengineer, projects engineer, and electrical supervisor.

The Synapse Partnership Ltd., 207A Ashley Rd., Hale,Cheshire, WA15 9SQ United Kingdom, E-mail:[email protected].

Biopharm Documentation


Biopharmaceutical ManufacturingDocumentationby Yong Wang

This articlepresents aprocess andoperationaldocumentationmodel for atypicalbiopharmaceuticalmanufacturingsystem. Eightimportantbiopharmaceuticalmanufacturingdocuments anddesigndocuments aredescribed andthe relationshipbetweendocuments isdiscussed.

Introduction

Biopharmaceutical processes are a se-ries of operations performed to makedrugs. These processes must complywith the current Good Manufacturing

Practice (cGMP) requirements, which are regu-lated by the US Food and Drug Administration(FDA). These processes, using biotechnologiessuch as fermentation, cell culture, recovery,and purification to produce drug bulks, are themost complicated processes in the pharmaceu-tical industry. Biopharmaceutical processes in-volve many different professional backgrounds,such as industrial microbiology, cell biology,chemistry, analytical chemistry, biochemistry,and chemical engineering.

A biopharmaceutical process is usually abatch process. It may take more than a monthto complete and it can involve up to 30 unitoperations. Automation is used inbiopharmaceutical companies extensively toincrease the manufacturing reliability and toreduce the number of the operating personnel.The automatic control systems in abiopharmaceutical manufacturing company are

Table A.Biopharmaceuticalprofessional’spercentage time indocumentation.

Profession Area Percentage of the Working Timein Documentation

Process Engineering 30-70%

Validation 50-90%

Automation 30-50%

Manufacturing Supervisor 30-70%

Operator 20-30%

Facility Engineering 20-30%

Maintenance Worker 10-20%

Quality Assurance 40-70%

Chemical Analysis 20-30%

Chemical Analysis Method Developer 30-40%

Management 30-50%

complicated because of the complexity of thebioprocesses.

Years of practices at biopharmaceuticalmanufacturing made the engineers in this areabelieve that during biopharmaceutical manu-facturing operations, manual operations mustbe added or combined with automatic opera-tions. The combination of manual and auto-matic operations makes the biopharmaceuticalprocess more complex.

Due to the complexity of biopharmaceuticalmanufacturing, the working ranges of eachbiopharmaceutical professional are narroweddown to small sections. Narrowing down work-ing ranges or professional ranges requires lessdiscipline, less training, and less experience tothe working professional. It helps the profes-sional become more focused and efficient.

On the other hand, many biopharmaceuticalprofessionals do not have a chance to see abroader picture of the whole biopharmaceuticalmanufacturing process. This can cause prob-lems because some of the roles in abiopharmaceutical company require knowledgeof the whole picture. To overcome this

problem, some biopharmaceuticalcompanies or the biophar-maceutical divisions of pharmaceu-tical companies, often encouragepeople to move from position to po-sition. This enables them to getmultiple discipline training, to be-come knowledgeable at a broaderpicture, and to have the ability toforesee something before it hap-pens.

However, it takes a long time forpeople to move from section to sec-tion to get the experiences andknowledge. It is even more difficultfor the professionals outside of thepharmaceutical manufacturing di-





vision to understand the details of the biopharmaceuticalmanufacturing. These professionals are the scientists/engi-neers who work in the R&D division of biopharmaceuticalcompanies: the process engineers, the validation engineers,and the automation engineers who work for engineeringfirms or consultant companies. It may be a good idea to letthese people know some details of the broad picture of how abiopharmaceutical company operates because they are in-volved in the various details of the biopharmaceutical manu-facturing.

In this article, a process and operational documentationmodel for a typical biopharmaceutical manufacturing systemwill be presented. Some of the important biopharmaceutical

manufacturing documents and design documents will bedescribed and the relationship between documents will bediscussed. The purpose of this article is to let people, who areinterested in biopharmaceutical manufacturing, understanda broad picture from the angle of process engineering throughprocess documentation activities.

Importance of DocumentationIn section 211.100(a) of the FDA document: 21 CFR Part 11(4-1-02 Edition)1, the FDA states, “There shall be writtenprocedures for production and process control designed toassure that the drug products have the identity, strength,quality, and purity they purport or are represented to pos-sess.” In section 211.188 of the same document, the FDAstates: “Batch production and control records shall be pre-pared for each batch of drug product produced and shallinclude complete information relating to the production andcontrol of each batch.” Here the FDA requires a licensedpharmaceutical manufacturing company not only completelyto record all the manufacturing information at the time of theperformance, but also to have all the manufacturing proce-dures written before the operations. These written processprocedures should promise the correct products to be made inthe specified quality, which the pharmaceutical manufactur-ing company addressed to FDA in license application. All thepharmaceutical manufacturing companies, including thebiopharmaceutical manufacturing companies, must complywith the documentation required to make bulk pharmaceuti-cals or pharmaceutical products.

Documentation is an important daily job for mostbiopharmaceutical manufacturing professionals. The pro-duction related activities should be documented according tocGMP and almost all the activities of the biopharmaceuticalmanufacturing professionals are production related. It isobserved that pharmaceutical professions spend huge effortsin document activities. Table A shows what percentage of theworking time a biopharmaceutical professional spends in thedocumentation related activities. The data in Table A isobserved or estimated by the author since there is no surveydata available in this aspect.

It is understandable that a high percentage of the totalbiopharmaceutical manufacturing salary cost is for the manu-facturing documentation activity. The biopharmaceuticalmanufacturing documentation is extremely costly in normalsituations.

It will cost more in abnormal situations. It is estimatedthat to a new drug, one day behind its marketing schedulemay cost a pharmaceutical company up to $1 million.2 Manyof the delays can be excused on documentation becausecertain important documents are not ready according to theschedules. Furthermore, biopharmaceutical manufacturingcompanies usually use relatively large batch scale. Eachbatch of a biopharmaceutical product may cost millions ofdollars. Mistakes in operations may result in quality prob-lems or losing a batch. It is not acceptable for abiopharmaceutical manufacturing company to risk qualityproblems or to lose a batch by malfunctioning.

Table of Contents

1. Introduction1.1 Purpose1.2 Scope1.3 References

2. Overview2.1 Functional Requirements

2.1.1 Security2.1.2 Structure and Documentation

3. Functions3.1 Operating Modes of the TANK

3.1.1 System Startup3.1.2 IDLE Mode3.1.3 RUN Mode3.1.4 SIP Mode3.1.5 CIP Mode3.1.6 CIP Parameters3.1.7 CIP Alarms

3.2 Control Loops3.2.1 Loop Status Assignments Available3.2.2 Loop Status Definitions3.2.3 Control Loop Descriptions

3.3 Discrete Outputs (Valve Matrix)3.4 I/O Lists3.5 Limit Switch Digital In and Digital Out

3.5.1 TA-30003.6 PLC File Designations

3.6.1 PLC Ladder File and Data Table Usage3.7 Inputs Provided for Non-Main Equipment3.8 Alarms/Interlocks

3.8.1 Alarm Types, Generation, and Acknowledgement3.8.2 Alarm Listing3.8.3 Interlocks

3.9 Manual Overrides/Forcing

4. Data Access4.1 Data from Tank to SCADA4.2 Data Input to SCADA by Operator4.3 Data Processing by SCADA4.4 Data Storage by SCADA4.5 Data Output by SCADA (Printed Reports)4.6 Data Displays at SCADA

5. Interfaces5.1 PLC Operator Interface

5.1.1 OIT Screen Map5.1.2 General OIT Screen Functions

5.2 Fault Tolerance5.3 User Entry Fields5.4 Alarms5.5 WIN911

Figure 1. Table of Contents of an FRS for a local control system.



One of the goals of biopharmaceutical manufacturingdocumentation is to reduce the risk of making mistakesduring operation. The method is to guide the well-trainedoperators through the manufacturing steps carefully usingthoroughly considered and tested manufacturing procedures.Biopharmaceutical manufacturing managers spend tremen-dous effort on documentation for this purpose.

Since biopharmaceutical manufacturing documentationis important and costly, studying and understandingbiopharmaceutical manufacturing documentation activities,and promoting the biopharmaceutical manufacturing docu-mentation efficiency will help professionals to reduce thepharmaceutical manufacturing cost. All levels of managersin a pharmaceutical manufacturing company are directlyinvolved in the biopharmaceutical documents’ drafting, re-viewing, approval, changing, and execution. Abiopharmaceutical manufacturing documentation level usu-ally directly reflects the biopharmaceutical manufacturingcompany’s management level.

Documentation is not a very pleasant process. People,when creating documents, must be fully concentrated. Phar-maceutical manufacturing documentation requires knowl-edgeable and skilled people since the documentation con-tents, as well as formats have to meet a manufacturingcompany’s quality standard. Documentation could also be anendless process. People can work on a document forever toimprove its quality. How to create a qualified document in alimited time span or in an efficient way is an art and mayinvolve some talent. Quality assurance personnel need tocontrol the documentation quality at a proper level. Projectmanagers should not underestimate the documentation ef-forts.

Bioprocess Engineering StrategyAs mentioned before, typical biopharmaceutical manufactur-ing involves a high degree of automation and also involvessome manual operating procedures that include prepara-tions or setups. There are two process engineering strategicreasons behind this fact.

Most biopharmaceutical manufacturing facilities are de-signed to make multiple products, and even a facility isdesigned for making single product, it is always expected thatother products may be produced in the facility in future.Adding manual preparation steps will make the operations ofthe manufacturing facilities more flexible. It is noticed thata bioprocess is composed of multiple operating procedures.Each operating procedure serves its own function. The differ-ences between bioprocesses can be expressed as whichfunctionalities are involved in certain order. It makes processengineering easier to divide a bioprocess into the operatingprocedures in their functionalities even in one unit operation.It is practical and beneficial to define and develop operatingprocedures according to the functionalities. These functionaloperating procedures, after being connected using automaticor manual procedures, form the bioprocess. Since it is easier,simpler, and more convenient to use manual operation proce-dures for transitions between functional operating proce-

dures, manual operating procedures are used vastly fortransition or connection purposes. Manual operations alsogive an automatic sequence a good pause point for automa-tion development, verification, validation, and monitoring.These functional operating procedures, which contain auto-mation sequences, after being optimized to comply with thecGMP requirements, form Standard Operating Procedures(SOPs). Breaking bioprocesses into functional operating pro-cedures makes it possible for many people working on a linearbioprocess at the same time. It also makes bioprocess changeseasy because usually a change only involves certain operat-ing procedure(s), there is no impact on the rest of the proce-dures.

There is another important reason to break down abioprocess into pieces. Any biopharmaceutical manufactur-ing facility involves the process phase and the cleaningphase. There is a requirement to segregate the process pipingsystems from the cleaning piping systems. Valves used forthe separation purpose are not considered as 100% reliable.System separation using valves may fail and cause severeresults. The best way for separating the piping systems is tohave a physical segregation between both piping systems.Transfer panels are used for the physical segregation be-tween piping systems. However, using a transfer panel re-quires manual preparations and set ups. Furthermore, op-eration on a transfer panel involves breaking a sealed andpotentially pressured process system. For safety concern andother reasons, some of the manual valves are added to processpiping systems to protect the operation personnel from ener-getic, chemical, and biological hazards during manual set upsor prevent potential important process piping leaks. This alsorequires manual operating procedures involved inbiopharmaceutical manufacturing.

Process DescriptionProcess description is the core bioprocess document to allbiopharmaceutical manufacturing companies. It may be calleddifferent names by different biopharmaceutical manufactur-ing companies. Process description describes under certainprocess conditions, how a pharmaceutical bulk or a drugproduct is produced. It gives all the process steps and proce-dures in chronological order and gives the process controlspecifications in the different process steps. It gives theprocess titer range and rough recovery rate for each processstep. It gives rough material balances. It gives the recipes ofall the media and buffers. It not only estimates the quantitiesof all the materials, such as chemicals, ingredients, andsolvents, involved in the process, but also gives all the qualitycontrol requirements of the raw materials. A full version of aprocess description may include the scientific background ofprocess theories and the related process procedures. It alsomay include the summary and the lessons learned fromprocess development. It may point out where the criticalmanufacturing steps are, and it may tell what will happen ifthe process controls are out of the specification ranges atthese critical steps. For example, in a pharmaceutical fer-mentation process, increase of fermentation temperature



from 28°C to 30°C will increase the ratio of an unwantedbyproduct, which has very similar chemical structure to thedrug product and which will cause purification problems. Inthis case, it is indicated in the process description thatalthough the optimum fermentation temperature controlrange was 28°C ± 0.5°C at the stage, once the fermentationtemperature reaches 28.6°C, an alarm should be triggered toremind the process engineers and operators to pay specialattention to prevent the fermentation temperature reaching29-30°C. Otherwise, the final impurity in the product willincrease up to 0.5%. A process description is usually draftedby the process development scientist/engineers. The readersof the document are the process related engineers, scientists,

and managers. A process description is the most confidentialbiopharmaceutical manufacturing document. Not every en-gineer/scientist in a biopharmaceutical manufacturing com-pany has the access to the full version of the documentbecause of confidentiality reasons. Many people may onlyaccess a part of it. A process description is used for the processengineers and the supervisors to understand the bioprocess.It also is used for developing the main operation documentand used for deciding the process control strategies, processcontrol parameters, and the process control ranges during thedesign stages. Usually a process description is drafted by thescientists and engineers from the process development divi-sion according to the process summary and the developmen-

Description Tag IDLE RUN SIP 1 SIP 2 SIP 3 SIP 4 SIP 5 SIP 6Mode Mode Prep T<100 T≥≥≥≥≥100 Hold T>100 100≤≤≤≤≤T

F E F E F E F E F E

Media-1 relay P-3001C X P X X X X X X X X X X X

Acid relay (pump) P-3001A X P X X X X X X X X X X X

Base relay (pump) P-3001B X P X X X X X X X X X X X

Agitator AG-3000 X P X O O O O O O O O O O

Exhaust FV-020 X O X O O X X X X X X O O

Vessel drain FV-023 X X X X O X O X O X O X X

CIP to sample valve FV-024 X X X X X X X X X X X X X

Acid port FV-025 X P X X O O O O O O O X X

Base port FV-026 X P X X O O O O O O O X X

Media-1 port FV-027 X P X X O O O O O O O X X

Jacket Blowdown FV-029 X X O X X X X X X X X X X

CIP Header Drain FV-034 X X X O O O O O O O O O O

Media-2/Inoc. port FV-035 X P X X O O O O O O O X X

Circ pump PU-3000 X O X X X X X X X O X O O

Heat TV-050 X P X X X X X X X X X X X

Cool TV-051 X P X X X X X X X O P P P

Glycol to seal HE TV-052 X O X X X X X X X O O O O

Jacket glycol empty TV-053 X X O X X X X X X X X X X

Jacket drain TV-054 X X X O O O O O O X X X X

Sparge port trap TV-056 X X X O X X O X P X O X X

Sparge filter drain TV-057 X X X O X O O O O O O X X

Sparge gas valve TV-058 X O X X X X X X X O O O O

Exhaust filter trap TV-059 X X X X X O O O O O O X X

Exhaust drain TV-060 X X X X X O O O O O O X X

Jacket steam TV-062 X X X O X P X P X X X X X

Jacket recirculation FV-063 X O X X X X X X X O O O O

Clean Steam to Header TV-066 X X X O O O O O O X X X X

CIP/SIP to Sparge TV-067 X X X O O O O O O X X X X

CIP/SIP to Overlay TV-071 X X X O O O O O O X X X X

Filter Drain TV-073 X X X O X O O O O O O X X

Key: X=Off-Closed O=On-Open NA=Not Applicable P=Pulse-Condition

Table B. Sample of a valve matrix.



tal batch records. Process engineers from the manufacturingdivision also may be involved in the activity. After pharma-ceutical production starts, the drug process will be continu-ally improved or updated by the manufacturing division. Theprocess engineer responsible for the bioprocess usually is theresponsible person to update the related process description.

Process NarrativeWhen a biopharmaceutical manufacturing company or apotential biopharmaceutical manufacturing company requestsan engineering company for a new manufacturing facilitydesign, a document called a process narrative needs to besupplied to the engineering company. The engineering com-pany will develop the process narrative into a Basis of Design(BOD) document through a process conceptual design.

A process narrative is a partial version of the processdescription. It gives as needed process information to anengineering company. It describes what is going to be made,and describes the process steps and procedures. It gives theprocess titer range, rough recovery rate for each processsteps, and some basic material data. It gives the target of theannual yield of the facility, minimum and maximum. Usuallythe scientific background of a bioprocess, the tricky part of theprocess, and process reasons may not be included in a processnarrative. Media or buffer recipes must be supplied to thedesign engineers for material balance calculations. Some-times, all the supplying raw materials may be shown asmaterial A, B, C, D, etc. in the recipe part of a user require-ment for confidential reasons. Another approach for keepingbioprocess secrets is to only give out the main ingredients inmedia or buffer recipes. The main ingredients in recipes arevery important for engineering calculation such as the mate-rial balance calculations. While the trace components of arecipe, especially a media recipe, play important confidentialroles in bioprocess. Without showing trace components in arecipe, a bioprocess secret can be kept because the tracecomponents can’t be predicted or estimated. The trace compo-nents in a media recipe are the trace amount of vitamins,biotins, metals, and salts.

Different companies may call the process narrative othernames, such as process description, design narrative, esti-mate narrative, or user requirement. An engineering com-pany may be asked to draft a process narrative by a potentialbiopharmaceutical company due to lack of process engineer-ing force. In this case, the potential biopharmaceutical com-pany needs to provide process development summaries to theengineering company.

Basis of Design (BOD) DocumentIn a conceptual design, the early phase of a facility design, theprocess engineers digest the information in the process nar-rative, make the process flow diagram, and do a series ofmaterial balance and heat balance calculations. The processengineers will lay out all the process requirements to alldifferent disciplines, such as architecture, HVAC, electrical,and environmental. The process engineers also will list themain process equipment and all the supporting utility equip-

ment. The architects will make building layouts to fit theprocess requirements. The project engineers will make a planfor the schedules, the milestones of the design and construc-tion, and capital investment in certain facility scale. Differ-ent scales of facility may be laid out for the pharmaceuticalcompany to review and to make decisions. After discussionsand modifications, the finalized conceptual design summarybecomes the Basis of Design (BOD) document for the facilitydesign project. Different companies may call the BOD a scopedocument. The scope of a BOD varies from company tocompany. Some companies’ BOD documents contain muchmore than others.

Some biopharmaceutical manufacturing companies, whichhave enough process engineering force, are able to do aconceptual design themselves. These companies sometimessupply a scope document, which includes a process narrative,to an engineering company for further designs.

Operating ProcedureWhen an engineering company submits a set of Process andInstrumentation Diagrams (P&IDs) to a pharmaceuticalmanufacturing company for review or further design, a set ofdocuments called operating procedures are submitted withthe set of P&IDs. Usually engineers divide thebiopharmaceutical facility into many process systems duringdesign stages according to unit operations. A process systemis an equipment concept of a unit operation. Each processsystem is composed of the main equipment, where an impor-tant step of a bioprocess is carried out, and its related piping.A process system can be illustrated in one or a few P&IDs.Usually one operating procedure is written for each operatingsystem. An operating procedure is a design document todescribe how a process system on the P&IDs is considered tobe operated by the design engineers. An operating procedureis composed of one or several functional operating proce-dures. All the manual operating procedures and the begin-ning step and the end step of an automatic operation aredescribed in an operating procedure in chronological order.

Operating procedures are for instrumentation engineers,control engineers, or automation engineers in engineeringcompanies to understand the process details. The processengineers or the supervisors of a manufacturing company useoperating procedures to develop their SOPs. They outline themain and important operating procedures involved in theprocess system. It is not a good idea to spend too much timeor add too many details into an operating procedure becauseit is drafted at the preliminary design stage. At this stage,many details have not been finalized or developed. There willbe many changes afterward.

Sometimes, because of time or money, process designengineers may only walk through the P&IDs with the profes-sionals from biopharmaceutical companies or in other disci-plines of the design company without writing the operatingprocedures.

Operating procedures can also be called different names,such as sequence of operations or operational outline, bydifferent companies.



Biopharmaceutical ManufacturingControl Systems

To describe biopharmaceutical manufacturing automatic con-trol system documents, it is necessary to briefly describe thebiopharmaceutical manufacturing control systems first.

All the biopharmaceutical manufacturing control systemscan be considered as either local control systems or centralcontrol systems.

Local control systems are located very close to the processsystems they control. Usually they are installed in the localcontrol cabinets of the process systems. Typically, one localcontrol system only controls one process system. The Pro-grammable Logic Controller (PLC) system is representativeof the local control systems. About 20 years ago, whenpharmaceutical manufacturing automation was at its earlystage, PLC was used for pharmaceutical manufacturingautomation control. PLC control system using ladder logicsas its programming bases. Other control systems, using otherprogramming bases, are developed these years as local con-trol systems. A local control system can communicate withSupervisory Control and Data Acquisition (SCADA) systems,or other control systems.

A central control system controls many process systemsremotely. It also can control process systems with localcontrol systems through the local control systems. A centralcontrol system is usually located in a central control roomwhere the automation engineers or the operators are workingin a manufacturing facility or a whole manufacturing plant.The best benefit of using a central control system is that itmakes interactions between two or more process systemsmuch easier. A Distributed Control System (DCS) is devel-oped as a central control system. It can handle many morecontrol devices and measuring instruments at the same timeremotely. It also supplies data storage function. There arecontrol modules imbedded into a DCS control system, whichmake programming easy by configuring the imbedded mod-ules. It improves the operation efficiency by providing opera-tors with visibility in multiple areas of a plant. One improvedcentral control system is very popular in the biopharmaceuticalmanufacturing industry. It is considered a scalable processcontrol system. The system uses Windows NT as its platform.The system communicates easier with a PC because theyhave the same platform. People consider it a more user-friendly system because they are more familiar with PC’s

Figure 2. Sample schedule of biopharmaceutical manufacturing documentation.



platform. The system can be integrated with other controlsystems, which adds other functions to the system easily. Thesystem makes it easier to track the system changes and tovalidate the system.

Local control systems are used for controlling such processsystems as centrifuges, in which fast signal responses arevery important. Local control systems also are used for somesophisticated and discrete process systems, such as fermen-tors and chromatography skids. Using which kind of controlsystem also is dependent on the management philosophy ofthe manufacturing facility. Sometimes, local control systemshave been used in a facility in which there are many processsystems. In this situation, all the process systems standseparately or are less integrated and more operational flex-ibility has been shown. On the other hand, a central controlsystem, due to the integration of process systems, showsbetter cooperation between process systems.

Whatever kind of control system is used for a pharmaceu-tical manufacturing facility, automation engineers requireprocess engineers to supply detailed process instructions andthe automatic process procedures, which are called automa-tion sequences, for implementing the control systems.

Functional Requirement Specification (FRS)and Detail Design Specification (DDS)

FRS and DDS are documents involved in pharmaceuticalfacility control systems and automation. They are drafted atdifferent design stages. Generally speaking, FRS specifiesthe process system requirements to automation. DDS ad-dresses the solutions to the specified pharmaceutical facilitycontrol systems. These documents specify the control systeminfrastructures, the automation sequences, various controlparameters, and the operating safety features. However,using either local control systems or central control systems,the documentation contents and formats, and documentationdevelopment pathways are different.

An FRS for a local control system specifies all the processrequirements. Figure 1 is an example of the Table of Contentsof an FRS for a local control system. In Figure 1, typicalcontents of an FRS for a PLC system have been shown. Thescope and the purpose of the control system have been intro-duced. Chapter 3 of the Table of Contents is very important. Inthis Chapter, important process requirements have been speci-fied, such as the operating modes, control loops, the Automa-tion Sequences, I/O lists, PLC file designations, alarm/inter-lock information, etc. The automation sequences in the FRSare shown as a valve matrix. A valve matrix is a table, in whichall the automatic valve positions or pump status in a processsystem are listed for all the process steps. Table B shows anexample of a valve matrix. Drafting an FRS for a local controlsystem needs efforts from both process engineers and theautomation or control engineers. Engineers start to draft anFRS at the end of the preliminary design stage. A DDS of a localcontrol system usually is an updated or finalized version ofFRS. There are not too many structure changes from an FRSupgrading to a DDS for a local control system. Usually, at theend of the detail design stage or at the beginning of the

commissioning stage, people change the documentation titlefrom FRS to DDS.

An FRS for a central control system is much different froman FRS for a local control system. An FRS of a central controlsystem contains the scope and the purpose of the controlsystem, the description of control loops, the automationsequences, and alarm/interlocks etc. Process engineers usu-ally draft the FRS for a central control system. These engi-neers’ main efforts, when drafting the FRS, is specifying theautomation sequences, alarm specifications, and the inter-locks. These automation sequences are shown as a descrip-tive format. Each complete piece of automation sequence iscalled a recipe or a code and is numbered or named. Recipeswork with the operating procedures to complete the modernbiopharmaceutical operations. DDS of a central control sys-tem is drafted by automation or control engineers in thedetailed design stage. In a DDS, the descriptive automationsequences are shown as the programming language format.The automation system, the system infrastructures, the I/Oaddresses, and the data recording systems are specified. ADDS finalizes the specifications, process controls, alarms,and the interlocks required by the FRS.

Although FRS and DDS documents are developed by thedesign engineers, after turned over to a biopharmaceuticalmanufacturing company, process engineers are responsiblefor updating the FRS whenever a control system is upgradedor modified. The automation or control engineers are respon-sible for updating the DDS whenever the control system isupgraded or modified by the process engineers in abiopharmaceutical manufacturing company.

Standard Operating Procedures (SOPs)Biopharmaceutical manufacturing SOPs are a group of writ-ten instructions for certain process function(s). An SOP is anaccurate, clear, succinct, and detailed list of operating proce-dures for operating personnel. For example, if an SOP de-scribes to open a manual valve, if the location of the valve isdifficult to find, the SOP might describe where the valve isexactly located. SOPs involve operating procedures with gen-eral functions, such as CIP or SIP operating procedures.Usually, SOPs are linear operating procedures and do notinvolve multiple processing choices. SOPs tell the operator toselect which recipe for the operation and how to select theautomation recipe on a computer terminal. SOPs includemanual operations and involve operating a computer key-board or pressing buttons on a computer terminal. SOPs aredeveloped based on the operating procedure. An SOP gives thesafety instructions. Usually, an SOP is drafted by a supervisorof operations, by a process engineer, or by a pharmaceuticalmanufacturing consultant during the process system commis-sioning. An SOP is usually a small operating procedure unit.SOPs also are used to train operating personnel who arefamiliar with standard or general operations.

Manufacturing Process Descriptive (MPD)The main manufacturing operation documents are calledbatch sheets or batch documents because biopharmaceutical



processes usually are operated in a batch mode. Today, thesame document is called Manufacturing Process Descriptive(MPD) by biopharmaceutical manufacturing companies. MPDis the main operating procedures which describes how tooperate manufacturing facilities to make a drug. MPD reachesthe same degree of operation details as do the SOPs. MPDinvolves operating procedures with specific functions. MPDinvolves the multiple choices’ operations. An MPD is devel-oped according to its process description. An MPD coversoperating procedures for many process systems. Sometimes,some of the operating procedures with general functions arerepeated for several times. One of the benefits of developingSOPs is the repeated part of the procedures can be written asSOPs. These SOPs are referred in an MPD instead of describ-ing them several times in an MPD.

MPD are involved with the process data recorded by theoperating personnel. For most of the bioprocesses, the pro-cess data will directly be recorded in the MPD according toinstructions. When a process batch frequency is high, sepa-rating the process data record book from an MPD could beconsidered. Managers or engineers divide MPDs according totheir operational functioning area. For example, MPDs aredivided as fermentation MPD, recovery MPD, purificationMPD, and media and buffer preparation MPD, etc. Usually,an MPD is drafted by a process engineer, a supervisor, or acombination of the two during the process system commis-sioning. The narrative detail degree of an MPD may varyfrom company to company. It takes two months to six months

to draft an MPD, which depends on how complicated andwhat degrees of details it reaches.

Unlike process descriptions, the MPDs will not explainwhy a bioprocess should be performed in a certain way.

Summary and DiscussionThis article introduced a number of biopharmaceuticalmanufacturing documents. Figure 2 shows a typicalbiopharmaceutical manufacturing documentation sched-ule. In a facility, a central control system is to be used as acontrol system for the whole facility. It is expected that itwill take two and a half years to establish this newbiopharmaceutical manufacturing facility through design-ing, construction, commissioning, and validation. In Figure2, the time lines of the documents, which are discussed inthis article, have been shown. Some other documentationactivities, such as FAT, IQ, OQ, and PQ, although they arenot discussed in this article, are shown for comparison of thedocumentation time frames. Colors are used for expressingthe responsible disciplines for their documents. In this way,the author hopes to give readers an overall picture ofbiopharmaceutical manufacturing documentation activi-ties. Figure 3 shows a documentation information flow pathto show the biophar-maceutical manufacturing documenta-tion process geographically. In Figure 3, only partial sec-tions of BOD and partial relationships between sectionshave been shown because a full discussion about BOD is notthe purpose of this article. Also, in Figure 3, dashed lines are

Figure 3. Typical biopharmaceutical manufacturing documentation information flow diagram.



used to specify typical responsible boundaries among R&Ddivisions, production divisions of a biopharmaceutical com-pany, and engineering companies. However, sometimeschanges of responsible boundaries have been seen. For ex-ample, a biopharmaceutical company can hire an engineer-ing firm to prepare SOPs.

Process descriptions, MPDs, SOPs, FRSs, and DDSs arecritical biopharmaceutical manufacturing documents. Pro-cess description is considered as the process laws or theprocess bible, which must be followed during manufacturing.SOPs are the block documents, which describe manual oper-ating procedures, describe the procedures of choosing a properautomation recipe, and describe the procedures to start andend a special automatic operation. An SOP may involve oneto several pieces of automation sequences. These pieces ofautomation sequences are referred as recipes. The details ofan automation sequence can be found in an FRS as a formatof process procedure and it can be found in a DDS as aprogramming format. An MPD describes bioprocess proce-dures as a whole and refers SOPs when they are needed. MPDfollows the process directions from the process description.An MPD is more like a process thread, which attaches theSOPs as the blocks on the thread to form a biopharmaceuticalprocess. Usually, SOPs cover the common process proce-dures, such as CIP, SIP, and typical operations. MPD coversrare process procedures, which form individual differentbioprocesses. Both DDSs and FRSs related to the SOPs orMPD in two language versions to reveal the automationdetails. If multiple bioprocesses are involved, there will bemultiple process descriptions and multiple MPDs. MultipleMPDs are more like multiple threads, which attach similarblocks, the SOPs, on them in different order and differentorientations. FRSs and DDSs typically may not be changed asthe related SOPs are not changed.

Operating procedures are not critical bioprocess docu-ments. However, their existence will make it easier for thepeople responsible for developing SOP because an operatingprocedure includes the main procedures of SOPs. Usually, anOperating Procedure covers the whole range of a processsystem including CIP, SIP, and typical operations. However,an SOP covers one of the functional procedures, such as CIP,or SIP, or one of the operations.

BOD is a main facility design document. It gives thebeginning points for a preliminary design or a detail design.The finalized FRS and DDS represent the design results.

Some problems about documentation preparation havebeen seen. One of the examples is people did not understandthe requirement of an operating procedure. Adding much moredetails to operating procedures were requested. Since bioprocessdetails were still changing at the time, all the related detailedoperating procedures had to be changed accordingly from timeto time, which wasted time. Another example is that some-times people neglect the important process engineering effortsin preparing FRSs. People thought FRSs were documents forcontrol systems. So, only control or automation engineers wererequested to complete FRSs although control or automationengineers had no problems to draft a “FRS” and specify the

hardware of the control systems. The systems would be pur-chased and installed on time. There would be problems at thetime to run and to test the control systems because the recipes,the control software, were not ready.

Although this article talks about documentation inbiopharmaceutical manufacturing, the pharmaceutical manu-facturing documentation activities are similar and simple.Different pharmaceutical companies may have developeddifferent manufacturing documentation systems because ofdifferent document development history. However, the ele-ments which build up the documentation system, and thecatch points of the whole documentation system, must be thesame or very similar. For example, the automation sequencescan be moved out of the FRS as an individual document afterreorganization of the FRS and DDS. Some company maycombine the operating procedure with the automation se-quence into one document. There is an example of differentdocumentation systems shown in another article.3 If readingit carefully, you will find all of the important manufacturingdocuments described in it can be found in this article; how-ever, they are called different names.

References1. U.S. Food and Drug Administration, 21 Code of Federal

Regulations Part 211 - Current Good ManufacturingPractice for Finished Pharmaceuticals (4-1-02 Edi-tion).

2. Meyer, MBA, RPh, Douglas, “Clinical Supply Chain,” Phar-maceutical Engineering, Vol. 22, No. 5, 2002, pp. 8-16.

3. Herrmann, Wiebke S., “Designing a Standardized SystemQualification Process,” Pharmaceutical Engineering,Vol. 22, No 5, 2002, pp. 83-94.

About the AuthorYong Wang is a Senior Process Engineerwith Jacobs Engineering Group Inc. His cur-rent job responsibilities involve the design ofbiopharmaceutical manufacturing facilities.He has 21 years of experience inbiopharmaceuticals and biotechnology. Hisvast experience includes process develop-ment, process scale-up, process transfer, com-

missioning, facility validation, process supervision, processmanagement, and process design. He has drafted manyprocess documents, such as process description, scope docu-ments, FAT, SAT, IQ/OQ/PQ, FRS, SOP, batch documents,operating procedures, and raw material specifications. Healso was involved in other documentation activities. Hegraduated from East China University of Chemical Technol-ogy with a BS in biochemical engineering and graduated fromthe University of Pennsylvania with an MS in chemicalengineering. He is a member of ISPE. Wang can be contactedby email: [email protected].

Jacobs Engineering Group, 3 Tower Bridge, 2 Ash St.,Conshohocken , PA 19428.

Process Validation


Process Validation AcceptanceCriteria for Solid Dosage Formsby Pramote Cholayudth

This articleintroduces anacceptancecriteria forprocessvalidation toassess theproductionbatch undervalidation interms of dosageuniformity.

Figure 1. Processvalidation sampling andtesting plan.

One of the most critical issues in processvalidation of solid dosage forms is prob-ably the acceptance criteria in eachmanufacturing step. Process valida-

tion is associated with appropriate samplingand testing with respect to sample size, samplenumber, sampling frequency, sampling loca-tion, sampling procedure, and testing methodfor each particular type of samples. In general,pharmaceutical processing steps, their corre-sponding qualifications, and their samplingand testing plan may be illustrated as in Fig-ure 1.

When conducting process validation of soliddosage forms, a series of extensive samplingand testing activities has to be performed inthe step of bulk mixing and bulk product pro-cessing (unit dosing) which significantly influ-ence the product uniformity.

• Bulk Mixing: a blend uniformity is tested fortablet granulation after final blending, pow-der mix for encapsulation after final blend-ing.

• Bulk Product Processing (Unit Dosing): massuniformity and content uniformity aretested, e.g., compressed tablets, capsules.

Bulk Mixing:Blend Uniformity (BU)

In 1999, the United States Food and DrugAdministration (FDA) recommended (actuallyproposed for public comments in a Draft Guid-ance for Industry, ANDA’s: Blend UniformityAnalysis) an acceptance limit of 90-110% of themean (x ± 10%) with a Relative Standard De-viation (RSD) of No More Than (NMT) 5.0% onabout 6 - 10 blend samples for ensuring theadequacy of the mixing of active ingredientprovided that the blend sample size is no morethan three times the dosage unit weight. Thiscriterion is intended to apply to those productsof potency less than 50 mg or products withcomposition of active ingredient less than 50%.The FDA later received many comments on theguidance from many sources, e.g., Pharmaceu-tical Research and Manufacturers of America(PhRMA), stating that the RSD criterion is notbased on scientific merit, i.e., not based onstatistical justification. Finally, the draft guid-ance was withdrawn from the Web site on May17, 2002 after the Product Quality ResearchInstitute (PQRI)’s Blend Uniformity WorkingGroup (BUWG) submitted to the FDA a pro-posal, Blend Uniformity Recommendation on“The Use of Stratified Sampling of Blend and



Process Validation


Dosage Units to Demonstrate Adequacy of Mix for PowderBlend.” On December 30, 2002, the final (revised) report onthe recommendation was resubmitted to the FDA for ap-proval (PQRI was appointed by the FDA for researchingscientific-based regulations, e.g., blend uniformity analysiswhich is under the responsibility of BUWG, a committeewithin the institute).

In the BUWG recommendation, the FDA’s blend unifor-mity acceptance criterion is still in use except that thenumber of samples is limited to not less than 10. To providean alternative of addressing the RSD limit problem, a way ofcalculating an appropriate RSD limit is introduced in thisarticle. Starting from the 90-110% limit and under the nor-mality assumption, the Z scores at the lower and upperpercentage points covering 95% of the area under the normalcurve between the lower and upper limits (90-110%) can becalculated as follows:

USL – µ LSL – µZ = __________ = __________

σ σ

WhereZ = Z score at 95% confidence (significant level with two

tail, α/2 = 0.025) = 1.96USL = upper specification limit = 110% Target Potency(TP)LSL = lower specification limit = 90% TPσ = population standard deviationµ = population mean = 100% TP

110 – 100σ = __________ = 5.10% TP

1.96

The corresponding sample Standard Deviation (S or SD) forn blend samples may be calculated using the conversionfactor derived by the following equation:

S2

χ21 – α, n – 1 = (n – 1) __________σ2

Whereχ21 – α, n – 1 = Chi square at 1-α confidence interval with

n-1 degrees of freedomn-1 = degree of freedomS2 = sample varianceσ2 = population variance

No. of No. ofSamples χ2

0.9, n – 1 [3] Fn SD Samples χ20.9, n – 1 [3] Fn SD

6 1.61031 1.76220 2.89 15 7.78954 1.34063 3.80

7 2.20413 1.64990 3.09 16 8.54675 1.32478 3.85

8 2.83311 1.57187 3.24 18 10.08518 1.29832 3.93

9 3.48954 1.51412 3.37 20 11.65091 1.27702 3.99

10 4.16816 1.46943 3.47 24 14.84795 1.24460 4.10

12 5.57779 1.40432 3.63 25 15.65868 1.23802 4.12

14 7.04150 1.35875 3.75 30 19.76774 1.21121 4.21

*Calculation is based on 90 % confidence level’s values to increase the SD’s to account for the inevitably biased blend sample data.

Table A. Factors for conversion between population and sample standard deviations (90% confidence level*).

After derivation, it can be expressed as follows:

n – 1σ = _____________ S = Fn × S

χ21 – α, n – 1

WhereFn = Conversion factor for sample size n

n – 1= _____________

χ21 – α, n – 1

To make the calculation more convenient, conversion factorsfor some frequently used numbers of blend samples areprovided in Table A.

The corresponding sample SD can be computed as follows:

σ = Fn × SS(SD) = σ/Fn=10 = 5.10/1.46943 = 3.47% TP

In summary, the protocol limits for blend uniformity for 10blend samples are:

• control limit (blend uniformity): mean ± 10% absolute;SD ≤ 3.47% TP

In general, the protocol limits for blend uniformity for n blendsamples are:

• control limit (blend uniformity): mean ± 10% absolute;SD ≤ 5.10/Fn % TP

We can see that “10% absolute” is introduced in the limit. Inthe BUWG final report, the limit has been slightly modifiedfrom “mean ± 10%” to “mean ± 10% absolute.” For example, ifthe mean is 99.7% TP, the corresponding limit is 89.7 –109.7% TP.

One fact associated with blend uniformity is sampling biasas one never takes true blend samples from the blend, i.e.,segregation occurs during thief sampling under the existingtechnology resulting in biased or deviated blend uniformitydata. Such segregation also occurs during sample handling toQC laboratory and subsequently weighing prior to assay. Thetwo latter cases could be overcome by assaying the entireblend sample of size 1-3 times (sometimes 5 or 10 times if 1-3 times is no more practical) the dosage unit weight. So blend

Process Validation


uniformity is affected significantly by the sampling biasrather than weighing error or analytical error. From a re-search paper written by Berman and Planchard, “BlendUniformity and Unit Dose Sampling” – “We believe that thiswas due to sampling bias, and as a result, the blend specimensthat were assayed were not representative of the population.…led to lower concentrations of drug in the samples than in thepopulation. Consequently, the RSDs of the samples werebiased on the high side since the sample means were biased onthe low side.” Such a lower potency is explained to have beencreated from electrostatic charges resulting in overfilling ofthe thief chamber with fine powder. And that is the reasonwhy 1) only the SD values, as above, are used instead of theRSDs, and 2) the “± 10% absolute” criterion is used to avoidthe biased control limit - Figure 2.

In the article, the Standard Deviation Prediction Interval(SDPI) method for calculation of critical (maximum) stan-dard deviation for blend uniformity was introduced. Thismethod was later referred in the PDA Technical Report No.25, Blend Uniformity Analysis: Validation and In-ProcessTesting in 1997. In the report, tabulated critical SDs areprovided. For example, if n = 10, the critical SD is 3.841 wherethe calculation employs the equation below:

S10Scr = ___________

F1 – α, 9, n – 1

Wheren = number of blend samplesScr = the critical standard deviation for blend uniformity

dataS10 = the upper bound of standard deviation for a sample

of dosage units of size 101 – α = confidence interval (e.g. = 0.9)F = the F statistic

The Scr, maximum SD of blend uniformity data, will ensure atleast 90% confidence that the content uniformity test resultwill pass the USP stage 1 criteria (n = 10), i.e., RSD not morethan 6%. So the upper bound of the standard deviation for afuture sample of dosage units (stage 1: n = 10) will be:

S10 = 0.06 × [target concentration] = 0.06 × 100= 6.0% LP

Therefore,

S10 6Scr = ________ = __________ = 3.841% TP

F0.9,9,9 2.44034

WhereF0.9,9,9 = 2.44034

Figure 2. Taking a blend sample from cubic blender.

No. of Samples χ20.95, n – 1 [3] Fn No. of Samples χ

20.95, n – 1 [3] Fn

10 3.32512 1.64520 40 25.69538 1.23198

20 10.11701 1.37041 50 33.93029 1.20172

30 17.70838 1.27970 60 42.33930 1.18047

Table B. Factors for conversion between population and sample standard deviations (95% confidence level).

The blend uniformity is sometimes tested on two (or evenmore) sets of samples. For example, when two sets of 10 blendsamples are taken at the blending time 10 and 15 minutesrespectively. The two test results will more effectively dem-onstrate the trend of blend uniformity rather than a single setof 20 samples’ results and also help to identify the time forblend optimality.

Bulk Processing (Unit Dosing): Massand Content Uniformity

1. Mass UniformityEstablishing the protocol limits for bulk products with re-spect to the control and RSD limits for the dosage unitsweights, or mass uniformity, is based on both the pilotproduction lot data and official limits. For example, if theaverage sample SD for pilot production batch is 1.5% ofnominal weight calculated from at least 20 samples, thestandard error of the mean is 1.5/√10 or 0.47 % of nominalweight where 10 is the sample size. The protocol control limitis established, according to control chart criterion, at ± 3σabout the nominal weight. So it is required to convert SD toσ using a conversion factor computed in the same manner asabove, but at a higher confidence level, 95%, as shown inTable B.

In the example, the population (lot) SD (σ) will be 1.5 ×1.64520 or 2.47% of nominal weight. So the control limit forindividual tablets will be 3 × 2.47 or 7.4% about the nominalweight and the control limit for average weight (n = 10) is 3× 0.47 × 1.64520 or 2.3% about the nominal weight. However,the ± 3σ range, or ± 3 SD × Fn, must not exceed the official(USP XIX) limits, i.e., for nominal weight not more than 130mg, the 3 SD × Fn is not more than 10% of the nominal weight,for weight between 130 and 325 mg, not more than 7.5%, and

Process Validation


Data\Lot No Lot 1 Lot 2 Lot 3 Limits

Blend Uniformity Data (% TP) - from 7 blend samples

Sample 1 95.44 99.02 98.67

Sample 2 95.64 97.05 96.39

Sample 3 99.60 99.17 97.07

Sample 4 94.54 98.88 100.40 90.00-110.00

Sample 5 98.19 95.16 99.02

Sample 6 96.00 95.48 98.85

Sample 7 97.34 95.51 99.82

Mean 96.68 97.18 98.60 -

SD 1.77 1.83 1.43 ≤≤≤≤≤ 3.1

Content Uniformity Data (% LP) - from 30 tablets

Minimum 95.46 94.06 98.32

Maximum 102.22 103.22 107.12

Mean 99.49 100.57 100.75 USP

SD 1.88 2.20 1.91

RSD (%) 1.89 2.18 1.90

TP = Target Potency; SD = (Sample) Standard Deviation

LP = Label Potency, RSD = Relative Standard Deviation = (SD/Mean) x 100

Table C. Blend and content uniformity validation data.

Figure 4. Normal distribution curve for content uniformity.

for weight not less than 325 mg, not more than 5%. If thedosage unit weight (nominal) in this example is 125 mg/unit(≤ 130 mg/unit), the established control limit is acceptable,i.e., more stringent than the official limit.

Setting the RSD limit is based on the capability assump-tion (using Cp ≥ 1.0) and the official limit for dosage unitweight. The maximum RSD for individual dosage units maybe calculated as follows:

USL – LSLCp = ____________

6σ

110 – 90σ = __________ = 3.33% nominal weight (NW) for

6 × 1.0 weight ≤ 130 mg/unit

SD n=10 = 3.33/1.64520 = 2.03% NW (→ RSD Limit(n=10): ≤ 2.03%)

SD n=20 = 3.33/1.37041 = 2.43% NW (→ RSD Limit(n=20): ≤ 2.43%)

107.5 – 92.5σ = ____________ = 2.50% (NW) for weight > 130 –

6 × 1.0 < 325 mg/unit

SDn=10 = 2.50/1.64520 = 1.52% NW (→ RSD Limit(n=10): ≤ 1.52%)

SDn=20 = 2.50/1.37041 = 1.82% NW (→ RSD Limit(n=20): ≤ 1.82%)

105 – 95σ = ____________ = 1.67% (NW) for weight ≥ 325 mg/

6 × 1.0 unit

SDn=10 = 1.67/1.64520 = 1.01% NW (→ RSD Limit(n=10): ≤ 1.01%)

Figure 3. Normal distribution curve for USP content uniformity.

Process Validation


SDn=20 = 1.67/1.37041 = 1.22% NW (→ RSD Limit(n=20): ≤ 1.22%)

In summary, the protocol limits in this example (sample size= 10) are:

• control limit (individual weight): 100 ± 7.4 or 92.6 –107.4% NW with an RSD ≤ 2.03

• control limit (average weight): 100 ± 2.3 or 97.7 – 102.3%NW

General expression for mass uniformity acceptance criteria:

• Individual weights for sample size n:- Control limit = 100 ± 3.SD.Fn % nominal weight- RSD ≤ X/(3. Fn) %

• Average weights for sample size n:- Control limit = 100 ± 3.(SD/√n).Fn % nominal weight

WhereSD = average SD calculated as % nominal weightFn = conversion factor for sample size nX = tolerance value in official limits e.g. 10% for nomi-

nal weight ≤ 130 mg/tablet provided that the 3.SD.Fn value is no more than the tolerance value X%.

n = Sample size

A successful capability study on the tablet compressionmachine, for example, should have been carried out prior toevaluation of the pilot batch data so that setting the protocollimit is accurate enough.

2. Content Uniformity (CU)The official limit of 85-115% Label Potency (LP) is deter-mined. In the USP content uniformity acceptance criteria,the sample RSD limits, i.e., NMT 6.0% for sample size 10 andNMT 7.8% for sample size 30, are based on the lot RSD ofNMT 10% [7,11]. One may calculate a corresponding RSD forsample size n using conversion factors in Table B as follows:

From the relationship,

σ = Fn × SIf σ = 10% LP (i.e. Lot RSD = 10% assuming the lot mean

equals 100% LP), Fn=10 = 1.64520SDn=10 = σ/Fn=10 = 10/1.64520 = 6.08% LP (See USP’s)If n = 30, Fn=30 = 1.27970SDn=30 = 10/1.27970 = 7.81% LP(See USP’s)If n = 60, Fn=60 = 1.18047SDn=60 = 10/1.18047 = 8.47% LP

Since a validation sample size is generally larger than thatspecified in the USP, e.g., PQRI requires at least 60 dosageunits in the first stage, the RSD (8.47%) corresponding toUSP’s acceptance criteria may be computed as above. But theUSP concept of lot RSD, i.e., 10%, provides only 86.64% of

dosage units of the entire lot falling within the range of 85-115% LP (see calculation below, Figure 3).

In establishing a more stringent RSD in a validationprotocol, the percentage of dosage units falling within thecontent uniformity range of 85-115% LP may be designed, forexample 99%. Then the corresponding lot RSD is calculatedand finally the sample RSD. To demonstrate how the lotpercentages and RSDs above have been derived, one shouldstart with calculation of the Z scores at lower and upper limitsas follows:

USL – µ LSL – µZ = __________ (or = __________ )

σ σ

WhereZ = Z score at 95% confidence (significant level,

α/2 = 0.025) = 1.96

Figure 5. Distribution curves for blend and content uniformity forthe three validation lots.

Process Validation


Figure 6. Distribution curves for blend uniformity and mass/blend/content uniformity.

USL = upper specification limit = 115% LPLSL = lower specification limit = 85% LPσ = population standard deviation = 10% LPµ = population mean = 100% LP

115 – 100Z = __________ = 1.5 (upper side)

10

So in the case of the two-tailed curve, the Z scores at 85 and115% LP are -1.5 (probability 0.0668) and 1.5 (probability0.9332). The percentage of the area between 85-115% LP is0.9332-0.0668 or 86.64% (for USP criteria).

For 99% acceptance criteria, imply that α/2 = (1-0.99)/2 =0.005, 1-α/2 = 0.995, the Z scores at 85 and 115% LP are –2.5758(α/2 = 0.005) and 2.5758(1-α/2 = 0.995) respectively.

The percentage of the area between 85-115% LP under thenormal curve is 0.995-0.005 or 99.00%.

To find the lot RSD is demonstrated as follows:

115 – 100σ = ____________ = 5.82% LP (→ Lot RSD = 5.82%)

2.5758

If n = 30, Fn=30 = 1.27970

SDn=30 = 5.82/1.27970 = 4.55% LP (→ Lot RSDn=30 = 4.55%)

If n = 60, Fn=60 = 1.18047

SDn=60 = 5.82/1.18047 = 4.93% LP (→ Lot RSDn=60 = 4.93%)

In summary, the protocol limits for content uniformity ofsample size n at conforming rate of P % of dosage units fallingwithin 85-115% LP range are:

• Control limit: 85-115% LP• RSD ≤ 15/(Fn × Z0.005P+0.5)

Suppose a protocol requires, for content uniformity, samplesize of 60 units at conforming rate of 99% (of dosage unitsfalling within 85 – 115% LP), the RSD limit will be 4.93%, ifthe results are that the mean of 60 units is 99.25% and RSDis 3.75%. One can calculate the actual conforming rate asfollows: σ = 3.75 × 99.25 × 1.18047/100 = 4.39, Z1 = (115-99.25)/4.39 = 3.58479 (probability 0.99983), and Z2 = (85-99.25)/4.39= -3.24338 (probability 0.00059). The actual conforming rateP = 0.99983-0.00059 = 0.9992 or 99.92% - Figure 4.

As discussed earlier, the blend uniformity data may bebiased due to several factors. Therefore, blend and contentuniformity data are often not correlated. The following is acomparison of blend uniformity and tablet content unifor-mity in terms of numerical data and graphical presentationsfrom the same batches showing how blend uniformity is oftennot predictive of the overall batch uniformity. Such data wererecently generated during a prospective validation of a tabletproduct containing 8% of active ingredient and compressedinto 125 mg/tablet where 7 blend samples and 30 tablets weretaken from each lot and witnessed by the author. The PQRIsampling plan was not issued yet at the time of execution. Astatus of sampling bias in the blend uniformity data may beobserved, i.e., most of the individual results are below 100%TP - Table C.

Those sample statistics in Table C may be estimatedthrough statistical methods to be the corresponding popula-tion (lot) parameters and presented as distribution curves inFigure 5 series. From the presentations, it is obvious that theBU curves for all the three lots are biased (deviated) on thesame side, i.e., always shift from the CU curves to the lowerside. From the BU curves, there is a tendency that individualblend samples, if taken in the future, may have the assayresult exceeding (below) the lower limit. All the CU curves arevery steep showing an excellent degree of meeting the speci-fications as their tails (lower and upper) lie far from the lowerand upper limits, i.e., 85 and 115% LP.

Process Validation


The magnitude of bias or deviation from the true value foreach individual result is not always the same so it is possiblethat the results do not meet the protocol acceptance criteria,either control limit or SD limit, while the content uniformityresults are excellent. In this case, it doesn’t mean that sucha validation trial fails, but it still passes provided that aninvestigation has to be undertaken and documented on thevalidation report.

3. Blend Uniformity Evaluated from Dosage UnitDataAn alternative way to assess the blend uniformity is recom-mended by PQRI under the concept that a tablet compressionmachine or capsule filling machine is an excellent tool for blendsampling, i.e., feeding blended granules into the machine diesand then compress or feeding blended powder into capsulesrespectively. When each dosage unit’s weight is checked priorto assaying, blend uniformity data, or weight-corrected data ascalled in the PQRI report, can be obtained in addition tocontent uniformity result using the same sample’s data. Theadvantages of using dosage unit samples instead of blendsamples are explained in the PQRI’s recommendation report,e.g., eliminating the blend sampling error, accounting forsegregation after blending, etc. Such blend uniformity datawill provide an assurance that the entire lot of blend is uniform(at the time of unit dosing step). The PQRI’s acceptance limitfor blend uniformity, using 60 dosage units, are between 90-110% TP for each sampling location mean with an RSD, for 60weight-corrected units, of NMT 6.0% which is corresponding tothe lot RSD of 7.08% using the same conversion criterion asabove. The lot RSD 7.08% will provide about 84.22% of the lotfalling within the range of 90-110% TP.

To demonstrate the calculation, Z = (110-100)/7.08 =1.4124 (upper side) so the Z scores at 90 and 110% LP are –1.4124 (probability 0.0789) and 1.4124 (probability 0.9211).The percentage of the area between 90-110% TP is 0.9211-0.0789 or 84.22%.

The advantage of blend uniformity evaluated from dosageunits (tablets) over that from the blend samples is its capabil-ity to demonstrate the overall batch uniformity as evidencedby the following example. At the time of final review of thisarticle, the author has witnessed a protocol execution of ahormone tablet product with active ingredient 0.75 mg per100 mg of tablet. The protocol was designed to fully follow thePQRI sampling plan (10 blend samples for blend uniformityand 60 tablets by stratified sampling for content uniformitytesting) and acceptance criteria. Fortunately, all three vali-dation batch results are available in time and can be summa-rized in graphical presentations in Figure 6. Here is anotherexample demonstrating that blend uniformity (blend sample)is not always predictive of batch uniformity.

From preliminary evaluation, the blend uniformity curves(red) generated from tablet data (weight-corrected data*) areat about the same location as the content uniformity curves(blue) while the blend uniformity (blend sample) curves (vio-let) have no repeated patterns, i.e., inconsistent in both loca-tion (mean) and spread (variability). Such, the blend sample

curves are always wider (longer tails) and create no predictablefeature of the overall batch uniformity. Each mass uniformitycurve (green) will demonstrate how adequate control of tabletcompression can provide an excellent content uniformity andalso demonstrate, with excellent mass uniformity, an interest-ing characteristic between the blend (weight-corrected) andcontent uniformity.

*Note: Weight-corrected data is content uniformity datatransformed into a nominal tablet weight database. Forexample, if 2 mg active ingredient in 120 mg per tablet isnominal, and the assay result for a tablet is 1.990 mg activeor 99.50% label claim (= (1.990/2) × 100) in 119 mg tabletweight, the weight corrected value is 99.50 × 120/119 =100.34% target potency.

ConclusionThe criteria for establishing validation protocol limits for apharmaceutical process are based on statistical techniques,e.g., using process capability or Z score under normalityassumption. The database used for establishing the protocollimits may be derived from two sources 1) official limits, e.g.,content uniformity, and 2) historical product data to estab-lish the natural and statistical limits, e.g., dosage unit weightsfrom the pilot production batch. The criteria related to officiallimits are slightly different in application in establishing theblend and content uniformity limits. For blend uniformity,the calculated lot sigma is converted to the correspondingsample standard deviation depending on the number of blendsamples. No RSD limit is established for blend uniformityunder the reason of sampling biased data. While the contentuniformity still follows the official control limit (85-115% LP),but employs a more stringent RSD instead of 6% dependingon the established conforming rate and the number of dosageunit samples. The criteria related to historical data arebeneficial for setting more stringent control limits for massuniformity, than the official limits. This article focuses on thequality attributes with respect to the active ingredient uni-formity; however, the other key attributes, e.g., size distribu-tion data, product assay, dissolution rate, etc., also are takeninto account in process validation.

To ensure that the established protocol limits are valid, itis recommended that a few pre-validation batches are pro-duced to have the key process parameters characterized priorto establishing the limits. A successful process validation isthe outcome of a successful process, i.e., a process with highaccuracy (the lot mean approaches target) and high precision(the lot standard deviation approaches zero). The samplingplan with respect to the sample size and sampling location isanother key issue for successful process validation as thelarger and more representative the samples, the less biasedresults.

References1. The United States Pharmacopeial Convention, Inc., United

States Pharmacopeia 24 ed. Philadelphia, PA: NationalPublishing.

Process Validation


2. The United States Pharmacopeial Convention, Inc., UnitedStates Pharmacopeia 19 ed. Philadelphia, PA: NationalPublishing.

3. Montgomery, D.C., 1991. Introduction to Statistical Qual-ity Control. 2d ed. New York: John Wiley & Sons.

4. Bolton, S., 1997. Pharmaceutical Statistics: Practical andClinical Applications, 3rd ed. New York: Marcel Dekker,Inc.

5. Chow, S.C., and Liu, J.P., 1995. Statistical Design andAnalysis in Pharmaceutical Science: Validation, ProcessControls, Practical and Clinical Applications, 3rd ed. NewYork: Marcel Dekker, Inc.

6. Berman, J., and Planchard, J.A., “Blend Uniformity andUnit Dose sampling,” Drug Development and IndustrialPharmacy, 21(11), 1257-1283(1995).

7. Mergen, G.J., “Evaluation and Validation of Blend Uni-formity: Approaches for Establishing Acceptance Crite-ria” Journal of Validation Technology, Volume 7, No. 2,February 2001.

8. Guideline on General Principles of Process Validation,Center for Drugs and Biologics and Center for Devicesand Radiological Health, Food and Drug Administration,May 1987.

9. PDA Technical Report No. 25, Blend Uniformity Analysis:Validation and In-Process Testing, Parenteral Drug Asso-ciation, 1997.

10. Final Report on Blend Uniformity Recommendations “TheUse of Stratified Sampling of Blend and Dosage Units toDemonstrate Adequacy of Mix for Powder Blends,” Decem-ber 30, 2002, Blend Uniformity Working Group (BUWG),Product Quality Research Institute (PQRI), www.pqri.org.

11. Email message dated June 11, 2001 from Dr. Larry Paul,General Policies and Requirements, Information andStandards Development, United States PharmacopeialConvention, regarding the source of RSD requirementsfor the USP content uniformity test, www.usp.org.

12. Process validation data for a tablet product 10 mg/tablet,April 2001.

13. Process validation data for a hormone tablet product 0.75mg/tablet, May 2003.

About the AuthorPramote Cholayudth is a Production Direc-tor for Biolab, Co., Ltd., a leading genericdrugs manufacturing company in Thailand.He is a guest speaker for Process Validation,held for industrial pharmacists, sponsored bythe FDA, in addition to being a guest lecturerfor GMP, Validation, and Special Problems inIndustrial Pharmacy in several universities.

He was previously a full time lecturer in the School of Phar-macy, Department of Industrial Pharmacy in a private univer-sity for four years (1998-2001). Prior to entering the academicarena, he spent 23 years in the pharmaceutical industry withBayer Laboratories (1974-1981) and OLIC (Thailand) Limited(1981-1997). Cholayudth is the author of Concepts and Prac-tices of Pharmaceutical Process Validation which has beenreviewed by the university’s academic committee. He can becontacted at: tel: +662-709-3121-2 ext. 3080, fax: +662-740-9586, or e-mail: [email protected].

Biolab, Co., Ltd., 625 Lane 7A, Bangpoo Industrial Estate,Muang, Samutprakarn, 10280 Thailand.

Electronic Documentation


Electronic Documentation -The Rewards of Information andProactive Implementationby Glenn Schulz and Gerhard Werling

This articlediscusses thebenefits ofportalapplications forcentralizingcriticalelectronic data,applicationrequirementsand essentialfeaturesdesigned tosupport 21 CFRcompliance.

Introduction

Anyone keeping an eye on the US Foodand Drug Administration (FDA) hasundoubtedly noticed the Agency’s con-tinual changes to the scope and appli-

cation of Part 11 of Title 21 of the Code ofFederal Regulations; Electronic Records; Elec-tronic Signatures (21 CFR Part 11). In Febru-ary 2003, the FDA issued a notice that it wouldre-examine 21 CFR Part 11 and make newrecommendations on requirements for valida-tion, audit trails, record retention, record copy-ing, and legacy systems. In the meantime,despite any changes the FDA is considering to21 CFR Part 11, adhering to internal documen-tation standards and implementing an inte-grated electronic document and change man-agement system can enhance your process,manufacturing, and help you address futureregulatory challenges.

Only time will tell what the final 21 CFRPart 11 regulations will look like, but thosepharmaceutical manufacturers who are headeddown the electronic documentation pathshouldn’t view the FDA’s re-engineering as areason to halt the process. For those of youmoving forward, this article will discuss thebenefits of portal applications for centralizingcritical electronic data, application require-ments and essential features designed to sup-port 21 CFR compliance. It also will touch onthe specific requirements for and benefits ofvalidation, audit trail, and security - three ofthe most critical issues surrounding regulatorycompliance. Lastly, it will outline practicalapplications of MES environments andworkflows.

Proactive Strategies forCentralizing Electronic Data

With the ever-growing list of manufacturinginformation systems comes an increase in ITand engineering support. Regardless of thebenefits achieved through validation, audittrails, record retention, etc., many manufactur-ers simply don’t have the time or the resourcesto manage and support so many individualregulatory-supporting applications.

In an effort to manage all these informationapplications, companies are starting to look atcentralizing plant-wide information, applica-tions and project files. Instead of multiple inde-pendently managed applications, portal-typeapplications now allow companies to funnel allthe information they need through one mainresource. Information can be gathered directlyfrom intelligent devices and through the soft-ware applications that manage and auditchange, prevent and predict failures, and verifyvalidation status of current projects.

Portal applications offer an array of ben-efits, including:

• Companies can control program usage acrossthe board.

• Companies can manage access to all files,projects, and products.

• Companies have a record of all (and mostcurrent) program versions running.

• Companies have a record of accepted con-figurations and programs running.





• Managers know when changes are made to all systemsdesignated to be tracked.

• Managers know who made changes or why they weremade.

• Companies can effectively run validated programs orregulated programs.

• Companies have records of all changes made to devices,applications, and project files being tracked.

• Operators can match programs and devices to safeguardproduction.

• Companies can prevent unwanted changes from occurringto devices, applications, and project files.

• Companies can restore previously used programs andcorrect invalid program changes in case of unauthorizedchanges.

Centralizing information does require an application thatprovides all the components needed to gather, store, manage,and report information from disparate sources. Key compo-nents include:

• Central server - the main server should manage services,databases, modules, and clients for the entire system. Thiscentral server requires fail-over provisioning to prevent asingle point of failure within the system.

• Event log - an event log is the centralized data repositoryand interface providing services that help store and dis-play warnings, errors, and informational messages. Eventlog functionality requires the use of an application thatcaptures events and pushes the information to the eventlog.

• Audit log - the audit log is a centralized data repository andinterface used to store and display edits occurring inmanufacturing products. Audit log functionality requiresthe use of an application that captures events and pushesthe information to the audit log.

• Service monitor - typically a server-based feature, a ser-vice monitor feature can allow users to monitor the stateof services running on any workstation or server.

• File management - a file management system allows usersto restrict and record the file usage. It should protectintellectual property and manage validated programs byrequiring users to check in and out of the system. It canmanage version history, making sure that changes madeto files are recorded and stored.

• Backup, recovery, and verification services - backup, re-covery, and verification services should support scheduleddata uploads and compares to devices and files. Someproducts offer built-in device support for controllers andother manufacturing hardware, but open driver supportalso should be available, providing third party device andsoftware support.

Figure 1. In an effort to manage critical information, companies are using portal-type applications to funnel plant-wide information throughone main resource.



Figure 2. Audit trail information should identify changes made tothe system, including the user and the reason for the change.

• Security - a security server or other security featuresshould be available to provide the ability to create rules ofusage of products and specific product actions based onuser and workstation names. In addition, some of the morerobust change management solutions offer additional lev-els of security that take into consideration not just who isaccessing the system, but where they’re accessing it fromand what they’re making changes to.

• Client - the client interface should allow users on net-worked workstations to access server functionality – tocontrol, display, search, and view capabilities for eventand audit databases, secure access to products and specificfeatures in products, and license check-in and out.

• License manager - product activation can be managedthrough a license manager to manage optimumconcurrency, restricted usage, historical usage informa-tion, license check-out status, and license location.

ValidationWith data centralized and a portal application in place,companies can focus on applying the features that will sup-port future regulatory requirements. The first is validation.The FDA’s Quality System regulation for the manufacturingof medical devices, which was published in the FederalRegister on October 7, 1996 and took effect on June 1, 19971

requires that “when computers or automated data processingsystems are used as part of production or the quality system,the manufacturer shall validate computer software for itsintended use according to an established protocol.”2

Validation of manufacturing software and systems im-plies that the actions taken by the user or the system itself (aswhen automated) have been executed with the proper legalauthority and formalities. Regardless of future FDA modifi-cations to validation requirements, validation of pharmaceu-tical manufacturing systems is still extremely important – asit is required by the predicate rules (e.g. Parts 210/211).Proving the accuracy of one’s manufacturing process, docu-

ment changes made to the system and identification ofinvalid or altered records is not just a matter of Federalregulation, it’s a matter of company liability.

To achieve computerized system validation and electronicrecords, a portal application should be able to automaticallyrun functions that support the on-going evaluation of themanufacturing process on a regular schedule. Routine - butcritical - operations like programmable logic controller up-loads, file backups, and compares can be set to run automati-cally at scheduled times. Important aspects of an effectivevalidation system include:

• the ability to run multiple events simultaneously, such asupload from several controllers at the same time

• automatically detect altered files and projects

• automatic notifications to operators when changes oralterations are detected

• built-in driver support for multiple controllers

• built-in support for third-party devices and products

• reporting capabilities

• security

A portal application such as Rockwell Automation’s elec-tronic maintenance documentation and change managementapplication is designed to centralize, manage, and maintaininformation for system validation. It can act as a central data-access point for audit trail, file management, and productlicense tracking information, making it a multi-purpose solu-tion for pharmaceutical manufacturers tracking and usingvarious data points - Figure 1.

Audit TrailsAnother important function for any pharmaceutical manu-facturer is the use of audit trails and the storage andretrievability of records created by a system. An audit trail isa secure, computer-generated, time-stamped report that in-dependently records the date and time of operator entries andactions that create, modify, or delete electronic records.Although they make up only part of 21 CFR regulations, audittrails offer substantial benefits to manufacturers, especiallypharmaceutical, chemical, food/beverage or consumer prod-ucts manufacturers – companies whose end products areused for human consumption. Most importantly, audit trailscan protect manufacturers in cases involving:

• Individual accountability - an individual’s actions aretracked in the audit trail making users personally ac-countable for their actions. This helps to deter users fromcircumventing security policies or making unauthorizedchanges to manufacturing systems. Even if users do act



outside policy or authorization, the actions captured in theaudit trail can help identify the user for accountability.

• Events reconstruction - in the case of unplanned down-time, machine failure, or other problems, audit trails canbe used to reconstruct events after the problem has oc-curred. Depending on the level of information collected,the extent and amount of damage occurring from anincident can be assessed by reviewing audit trails ofsystem activity to pinpoint how, when, and why the inci-dent occurred.

• Problem monitoring - Audit trails can be used as real-time tools to monitor manufacturing processes and/orproblems as they occur. Based on the information beinggathered through the manufacturing process, real timemonitoring can help detect process inconsistencies, ma-chine failure, over-utilization of system resources, or en-ergy outages.

• Intrusion detection - Intrusion detection refers to theprocess of identifying any efforts to penetrate a manufac-turing or enterprise computer system and gain unautho-rized access. Often, this access can originate within thecompany although external security breach attempts alsoare common. Audit trails can only help in intrusion detec-tion if they record appropriate security events.3

Despite the overwhelming benefits of audit trails, the abilityto effectively archive and retrieve data in a reliable, secureform is where many current systems fail to deliver. In thepharmaceutical industry, audit trail reports should be gener-ated for all user actions on all manufacturing devices andsystems. At the very least, audit reports must identify changesmade, the user, and the reason for the change. Audit recordsentered into an audit trail are typically identified with thefollowing: (Figure 2)

• time stamp recording the transaction time

• the kind of transaction (create, delete, modify)

• affected field name

• old value of the field

• new value of the field

• identification of the user who performed the transaction

Additionally, the following information can be added for eachdata change:

• the reason for the change, whenever appropriate

• an electronic signature, whenever appropriate

Once made, audit trail records and the data they report mustbe protected from alteration or deletion, and informationowners/managers should be able to easily identify unautho-rized changes.

SecurityBecause security is such an important aspect of electronicrecord keeping, all electronic records must be stored in asecure location and format. Electronic records are often keptin a relational database, such as the Microsoft SQL database.However, keep in mind that database tables should be lockedwithin the database, and access to tables restricted throughsecure layers. Some products use Microsoft Windows andMicrosoft SQL security. An effective validation and audittrail system also should prevent end users from modifyingrecords, further protecting the company against liabilitiesdue to possible information sabotage. For information report-ing, users can often choose from a variety of reporting toolsincluding reports within validation system software or sepa-rate tools, such as Microsoft Access, Microsoft SQL Servertools, or Crystal Reports. Keep in mind that these separatereporting tools are not 21 CFR compliant, as data can bemanipulated by the user. Only reports generated from archiveddata and protected from end user intervention or manipula-tion will adhere to 21 CFR requirements.

Applying Electronic Documentationin the Facility

Once you understand the history and benefits of an inte-grated electronic document and change management system,how can it be effectively implemented? Data within a phar-maceutical company can be referenced as Standard Operat-ing Procedures (SOPs). SOPs, which come in all types of formsand varieties, including bills of material and productionprocedures, need to be version-controlled. Today’s advancedManufacturing Execution Systems (MES) allow SOPs to bemanaged electronically. SOPs are usually written with aMicrosoft Word-compatible editor, and are created in threecommon ways:

• start with an existing SOP template (often called a MasterSOP), and make the necessary additions

• make a copy of an existing SOP and edit accordingly

• build an SOP from scratch

Word-compatible editors allow SOP developers to work in a“normal” office environment and use common editing fea-tures, while at the same time have the necessary document(version) control.

SOP approval requires the MES system to have a docu-ment workflow that is well-defined based on the company’spolicies. These rules can be effectively enforced through theMES system using its version control features. The systemsmay use a version graph function that reflects the approval



workflow assigned to each SOP object according to companyand GMP rules. In addition, electronic signatures can beattached to each workflow step to assure proper authoriza-tion.

For example, an employee in quality assurance can createa new SOP (from a template, an existing SOP, or scratch) andthen initiate the approval review. The type of review is ofcourse dependent on the type of SOP with more significantSOPs (such as those for master validation plans) requiringhigher-level approvals. A draft SOP might first go through apeer review, then quality management, then departmentalmanagement, and finally plant management if required.Each review (and in many cases corresponding electronicsignature) is documented in the version history. The systemenforces that only proper organizational roles may be in-volved in this workflow, thereby ensuring compliance tocompany procedures.

An SOP within the MES system also can be hyperlinked toa respective step within a production procedure. In this way,each employee can always have direct access to the validversion of the SOP that relates to the step they are executing.By double-clicking on the screen, the employee can quicklybring up the SOP – a useful tool also for in-process controlactivities. More advanced MES systems also allow users toquery the system to find only deviations from the standard inan executed batch. As such, the quality management group inan enterprise may focus attention on these deviations. Ver-sion control features allow any data objects to be controlled bya version graph similar to the one used for SOPs – trackingcreation, review, approval, valid use, and archiving.

It’s often advisable for an enterprise to perform a docu-mented risk analysis of required audit trail data to stream-line management. In a risk analysis, every relevant datarecord created or modified by the system receives an assess-ment of its GxP criticality – more specifically, you need to askif the data has a direct impact on the product quality orquality documentation. Moreover, only data that can bechanged by an operator via the normal user interface need tobe subject to audit trail. The change management and MESsystems should be set up to capture this key audit trail dataand make management of that data much more efficient.

ConclusionDespite the uncertain future of FDA regulations on electronicrecords and signatures, it seems clear that the ROI benefitsof integrated electronic documentation and change manage-ment are enough to propel its advancement in pharmaceuti-cal manufacturing. At the very least, companies get a headstart on some of the regulatory compliance mandates thatwill likely resurface within the year. At the most, companiescan start paving a path to collaboration and informationsharing, improve and increase workflow efficiency, and ulti-mately find cost-savings through information.

In addition, companies need to recognize that even moresignificant benefits can be derived from weaving electronic

documentation into the bigger disciplines of automatic policyenforcement and overall maintenance management. There-fore, looking at electronic documentation as part of a broadautomated policy enforcement strategy can compound itsbenefits. This broader abstraction calls for the ability tocontrol, monitor, and enforce activities through the intersec-tion of who (user and group), where (the current locale of theuser), and what (the source and target). This simply wouldn’tbe possible using manual methods.

References1. FDA 21 CFR Parts 808, 812, and 820,” October 7, 1996.

http://www.fda.gov/cdrh/fr1007ap.pdf.

2. FDA General Principles of Software Validation; FinalGuidance for Industry and FDA Staff,” January 11, 2002.http://www.fda.gov/cdrh/comp/guidance/938.pdf.

3. Rajeev Gopalakrishna, “Audit Trails,” April 2000. http://www.cerias.purdue.edu/homes/rgk/at.html#intro.

About the AuthorsGlenn Schulz became Director of GlobalBusiness Development for Rockwell Auto-mation in 2003. He is responsible for newproduct development, business alliances, andstrategic acquisitions under the RockwellSoftware brand. Schulz began his career withRockwell Automation in 1995 as a seniorreliability engineer. In 1996, he was pro-

moted to supervisor, product development. Schulz becamemanager, Product and Business Development, in 1997 andheld this position until his current assignment. Schulz earneda BS in electrical engineering from the University of Wiscon-sin-Madison. He be contacted at tel: 1/414-328-2150 or email:[email protected].

Dr. Gerhard Werling is the Director ofQuality Management and Validation Ser-vices for Propack Data GmbH, a business ofRockwell Automation. Head of his depart-ment since 1994, Dr. Werling acts as Propack’squality representative and manages valida-tion projects, consulting, and services forcomputerized systems. Dr. Werling is active

in the GMA/NAMUR working group “Validation of ProcessControl Systems” and is a member of both PDA and the ISPEGAMP Forum. Dr. Werling studied computer science andmechanical engineering at Karlsruhe University in Ger-many, and attained his doctorate at the Institute of ProcessControl Systems and Robotics of the University of Karlsruhe.He can be contacted at tel: +49-(0)721-9650-835 or email:[email protected].

Rockwell Automation, 1201 S. Second St., Milwaukee, WI53204.

Effective SOPs


Effective Standard OperatingProcedures in a RegulatoryEnvironmentby Erica Charlton

This articlediscusses theimportance ofStandardOperatingProcedures(SOPs) in aregulatoryenvironment andmakessuggestions asto how tocreate, maintain,and circulatethesedocuments in amannercompliant toworldwideregulatoryagencies.

Introduction

Regulatory agencies the world over, andspecifically the US FDA, require theexistence of a document repository todemonstrate that procedures and pro-

cesses in Life Science Manufacturing facilitiesare in place and followed by all personnel. If aprocedure or process directly affects a product,then it should be outlined in a current andapproved document. Creating and maintain-ing a collection of these does not have to be amajor headache if it is approached in a logicalmanner as shown in some of these basic guide-lines.

Standard Operating Procedures are the ex-plicit written description of a Production, Qual-ity Assurance, Materials Management, Admin-istration, Documentation, or Engineering op-eration performed by personnel in a GMP envi-ronment. An ‘operation’ is an activity whichmay affect the product conformance to specifi-cations or regulations. The SOP defines theessential steps, their sequence, and precau-tions necessary to uniformly repeat perfor-mances of the operation.

FDA regulations require that procedures bedocumented, which will be audited during anFDA inspection. There are no precise instruc-tions provided by the FDA as to how SOPs arewritten, stored, circulated, approved etc. It isthe discretion of the regulated company howSOPs are handled. In an audit, the FDA issimply looking for the existence of these proce-dures and that the SOP and the associatedpersonnel training maintains and strengthenscGMP systems, processes, and procedures.

From 21 CFR Part 211.100 (Current GoodManufacturing Practice for Finished Pharma-ceuticals, Subpart F; Production and ProcessControls, Written procedures; deviations):

(a)There shall be written procedures for pro-duction and process control designed to as-sure that the drug products have the iden-tity, strength, quality, and purity they pur-port or are represented to possess.

Such procedures shall include all requirementsin this subpart. These written procedures, in-cluding any changes, shall be drafted, reviewed,and approved by the appropriate organizationalunits and reviewed and approved by the qualitycontrol unit.

Initiation of an SOPThe following are several events that can ini-tiate the requirement for a new SOP: a newpiece of equipment, a Corrective Action Preven-tative Action (CAPA), a Deviation report, aValidation requirement, a company policychange, or a change in an affiliated document.

Writing SOPsWhen creating a new SOP, the ‘documentationdepartment’ (or the company’s equivalent) is-sues a unique number to the author. Logically,the best author for an SOP is a qualified indi-vidual who may or may not be the same indi-vidual executing the procedure. At the mini-mum, the SOP should be reviewed by someonewho performs the procedure so they may havetheir input with regard to the accuracy.

It is important that the wording in the SOPcontent be clear and concise. Minimize theopportunity for discrepancies and avoid flow-ery, descriptive text - just get to the point. It’snot a literary contest.

It’s safe to assume that there are probably asmany variations in SOP format and chosenheadings as there are companies using them.Some may include more or less and the chrono-



Effective SOPs


readers may require to fully comprehend the procedure.Figures should be numbered and titled appropriately.

Training - although it is up to the author, some companiesrequire a short collection of test questions in order thatdocumented training systems and the SOPs can be linked.Answers are found in the body of the SOP and if the traineehas read the procedure thoroughly, they are not difficult toanswer. Multiple choice, true/false questions are quick anduncomplicated. The test answers can be completed by thereader/trainee at the time of training to test the reader’sability to comprehend the procedure. This is a satisfactorymethod of creating a measurable training record for employ-ees if a company chooses this method.

Miscellaneous - Some companies may include a ‘Frequency’heading if it is a cleaning procedure or maintenance proce-dure to indicate when the procedure is required.

The SOP FormatEach company develops their own in-house method for thecategorization of SOPs and the format of the SOP body, butthe style should be consistent across all departments. Forcompanies with a large index of SOPs, they will find it usefulto group the documents by department, for example: allQuality Assurance SOPs are grouped under a common prefixcode, the same for all Manufacturing SOPs, or Administra-tive SOPs. The categorization can be further defined withsome form of a logical alphanumeric code, again, under thediscretion of the company.

The SOP template should include somewhere; the com-pany name, the current date, and a statement with some-thing to the effect that: “In printed form, this document is onlyvalid on the date shown.” Which is a method of identifyinguncontrolled documents. This will be expanded upon later inthis article.

An SOP cover page should include; the title of the SOP, itsunique number, the effective date, the author (by job title),the approver(s) (by job title), current version number, revi-sion numbers, and the revision history if applicable.

Within the body of the SOP, it is effective to create anumbering scheme for each heading that can be a drill downsequence in each section. For example,

1.0 TitleSubtitle

2.0 Title2.1 Subtitle2.2 Subtitle

2.2.1 Subtext2.2.1.1 etc.

Of course, the tab settings are optional, but it can create a tidylooking appearance.

logical order can vary, but the following is a list of headingsfrom SOP templates in many Life Sciences companies:

Purpose - this is a brief statement of one or two sentencesstating the reason for the SOP. It may read something like:“The purpose of this SOP is to outline the procedure for thecare and maintenance of instrument x.”

Scope - By writing the SOP, what do you hope to accomplish?What does this procedure apply to? The author states in thescope how much of the procedure is outlined in this SOP andwhat is not covered.

Responsibility and Authority - here the author indicateswho, by position or title, should be responsible for learningand executing the procedure. This helps to determine whoshould be trained on the procedure.

Environmental Issues - this section is sometimes includedin SOPs where there is the use of chemicals that could posean environmental threat. Here it is appropriate for alertingthe reader about the seriousness of chemical spills, harmfulgas releases, and how to handle them. It’s not necessary torewrite the entire procedure for handling chemical spills. Ifthere is a potential for chemical spills, there should be aseparate SOP for the proper clean up. The author may referspecifically to this SOP by number in this section or under theAssociated Documentation heading.

Safety - this section alerts the reader if extra care is neededwhen executing the procedure and what personal protectiveequipment is required, if any.

Associated Documentation - here the author cites instru-ment manuals, refers to additional SOPs or in-house docu-mentation, corporate procedures, or additional documentscited in the body of the SOP which might overlap or enhancethe information included in the SOP.

Definitions - these are simple word definitions for termsthat the author feels require clarification. Acronyms areincluded in this section, if applicable.

General Outline - the general outline can be used to indexthe text in the procedure usually by the main headings. It’smost helpful in long SOPs where the readers are looking forspecific information. Using the General Outline they can findinformation quickly.

Procedure - this is written for use as a training tool. This isthe step-by-step sequence of events. Additional equipment,tools, or chemicals are referred to by name. The author shouldidentify additional documentation and contact positions ifapplicable.

Appendices - these are attached diagrams or separate charts,tables, and or tools that enhance the text portion and that

Effective SOPs


Approving SOPsWhen a draft SOP is completed, it is circulated to the appropri-ate personnel for review. Approvers review to ensure thedocument properly reflects the procedure. It’s possible (ormaybe probable in some organizations) that discrepanciescould cause the document to cycle around more than once foradjustments before a final copy is ready for approval. Hope-fully, this can be addressed quickly to keep the process moving.

Be aware if any changes impact another part of thatparticular (or another) SOP and address those changes im-mediately.

The Document Manager (or similar position), from theQuality Assurance department, completes the final review ofthe SOPs. This person applies the effective date and launchesthe SOP from the ‘draft’ to ‘approved’ status and then live.Training for appropriate personnel should be arranged shortlythereafter.

Uncontrolled DocumentsMany facilities store a complete collection of bound hardcopies of SOPs at a specified location in their facility. It isimportant to note that any printed copies outside that loca-tion are classified as uncontrolled documents. The problemthe FDA sees with hard copies is the possibility that any givenSOP could be under revision (unknown to all personnel), andso the hard copy is no longer the current and valid copy.

Only current and approved versions of SOPs should beaccessible by personnel. Draft versions or those under revi-sion should be only accessible to authors and reviewers, andthey are launched into the live system only when approved.

SOP MaintenanceCompanies permitting hard copy collections of SOPs willhave an SOP on ‘How to Write SOPs.’ This would be an authorguideline and describe the information that should be listedunder each section heading, how versions are addressed, howthe SOPs are circulated for approval and launched into thelive system, how to eliminate the risk of uncontrolled docu-ments by including a print date on the hard copy SOP and theclause ‘valid only on date printed.’

An electronic document management system makes themost sense for writing, circulating, and maintaining the SOPcollection, and like any computer system that affects cGMP,it must be capable of audit and validation.

At a minimum, SOPs should be reviewed once every twoyears. Updates or revisions are completed as needed and thetraining-tracking database can flag appropriate personnelfor timely training/retraining. It may not always be theoriginal author available to review, but it should be someonewho performs the procedure described.

A revised SOP is circulated for approval and posted on thelive system in the same manner as an original copy, but thenumbering scheme for the document must somewhere indi-cate that it is the subsequent version of the original. A noteunder the ‘Revision History’ heading should briefly summa-rize the changes made, by whom, the date, and the previousSOP version number that it replaces.

Old versions of SOPs should be archived for an audit trail,but should not be accessible in the live system as they areUncontrolled Documents when out of date. An FDA auditorcan request proof of existence of previous versions of anydocuments and proof that the company can identify anychanges made through the history of the document’s exist-ence.

When SOPs become obsolete, they must be removed fromthe live system. In some cases, the company chooses anumbering system with almost an infinite number of possi-bilities whereby a number would never be reused. In othercases, the SOP number is retired for a predetermined amountof time, after which retired SOP numbers can be reissued fora new SOP. The retirement period is determined by thecompany policy.

The document management department is responsible forkeeping the SOP library in control in that there aren’tduplicate procedures being written by different people at thesame time. To minimize the library, it is wise to compilemultiple short procedures into a larger SOP if they arerelated. Be reasonable when creating SOPs. Decide whichprocedures warrant an official document. If the procedureaffects the product in any way, then it is a GMP issue andshould have an SOP to accompany it. An electronic documentmanagement system should be able to flag SOPs that are upfor review and by whom.

There are dozens of commercially available applicationsthat will effectively maintain SOP documents electronically.A search on the Web will yield many applications with thesame basic components. Choosing one depends on the fea-tures and functions required by the company.

Training PersonnelThe FDA requires that all personnel must have instanta-neous access to approved SOPs in their work environmentwhether in electronic form or a controlled hard copy version,so that they may refer to the steps of any procedure at anytime.

From 21 CFR Part 820.25, Quality System Regulations,Subpart B; Quality System Requirements, Personnel:

(a)General. Each manufacturer shall have sufficient person-nel with the necessary education, background, training,and experience to assure that all activities required by thispart are correctly performed.

(b)Training. Each manufacturer shall establish proceduresfor identifying training needs and ensure that all person-nel are trained to adequately perform their assignedresponsibilities. Training shall be documented.

And, also from 21 CFR Part 211.25, Current Good Manufac-turing Practice for Finished Pharmaceuticals, Subpart B;Organization and Personnel, Personnel Qualifications:

(a)Each person engaged in the manufacture, processing,packing, or holding of a drug product shall have education,training, and experience, or any combination thereof, to

Effective SOPs


enable that person to perform the assigned functions.Training shall be in the particular operations that theemployee performs and in current Good ManufacturingPractice (cGMP) (including the current Good Manufactur-ing Practice (cGMP) regulations in this chapter and writ-ten procedures required by these regulations) as theyrelate to the employee’s functions. Training in currentgood manufacturing practice shall be conducted by quali-fied individuals on a continuing basis and with sufficientfrequency to assure that employees remain familiar withcGMP requirements applicable to them.

In a GMP environment, the simple existence of an SOP doesnot indicate (to the FDA) that personnel are trained on theprocedure and therefore qualified to do his/her job. Thereshould be a system in place by which an employee signs anddates a document indicating they have indeed read andunderstand the procedure. This document becomes part ofthe training file.

Employee retraining is not required if the changes to anexisting SOP have no effect on the on-going procedure.

Some companies opt for a more advanced method ofensuring that employees have absorbed information includedin the SOP that enables them to perform the job correctly,consistently, and safely. The training effectiveness becomesmeasurable. For example, test questions whereby the correctanswers create a score. The training system must be capableof audit and validation. This is why it is advisable, in thissituation, to link the SOP document system with a trainingtracking system. Training tracking applications can be quitesophisticated and contain lists of students, courses, sched-ules, instructors, classrooms, student information, coursehistory, and course results. They are used for schedulingtraining and maintaining employee training files.

A training tracker application operates on a pyramid ofrequirements and must allow for SOPs to be linked to jobs,jobs must be defined in terms of skills, and skills must bedefined in terms of teachable elements. This kind of break-down leaves little room for ambiguity in employee’s skill sets.

In addition to the drill down pyramid structure, anotheradvantage to an electronic training tracking application is tocreate consistency across an organization as to when and howpersonnel are trained. The system administrator of the sys-tem can flag user profiles to indicate when retraining is due,arrange for training, and maintain results of training. WhenSOPs are revised, retraining for appropriate personnel isrequired, and the training tracking software should be able tonotify of this if they are integrated systems.

The administrator of the training tracking software is oftenthe company’s training coordinator. Within the application,they are able to assign SOP reading lists to all employees,accompanied with an expected completion date. Personnel willaccess the live SOP systems to read their required SOPs andcomplete the training questions for each. Answers to the testquestions can be recorded electronically or submitted as a hardcopy, and forwarded to the training coordinator for review. Thecoordinator updates the employee-training file.

There are dozens of software applications which can effec-tively manage employee-training records and an internetsearch will yield many possible choices.

In conclusion, SOPs are a necessity. Follow these rules ina GMP environment:

• What requires an SOP?A procedure that affects the products’ conformance tospecifications or regulations must have the steps capturedin an SOP document.

• When are SOPs written?SOPs should be written when there is need for a newprocedure. They should be reviewed periodically and re-vised as needed.

• Where are SOPs stored?The most efficient method for storage and revision is anelectronic document management system which makesthem accessible by all employees. There also may be hardcopy versions that are strictly maintained.

• Who writes SOPs?The experts are responsible for writing the SOP, in otherwords, those who perform the procedure. The QualityDepartment approves SOPs.

• Why SOPs?SOPs are necessary because the FDA says so. Ultimately,they are for the protection of everyone: employees, theemployer, the public. It’s just plain smart business.

References1. American FDA, Code of Federal Regulations 21 CFR Part

211.100, Current Good Manufacturing Practice for Fin-ished Pharmaceuticals, Subpart F, April 1, 2002.

2. American FDA, Code of Federal Regulations 21 CFR Part820.25, Quality System Regulations, Subpart B, QualitySystem Requirements, April 1, 2002.

3. American FDA, Code of Federal Regulations 21 CFR Part211.25, Current Good Manufacturing Practice for Fin-ished Pharmaceuticals, Subpart B, April 1, 2002.

About the AuthorErica Charlton is the Marketing Represen-tative at PENSA Technology Solutions Inc.,a privately owned Canadian company whichis uniquely qualified to plan, design, imple-ment, and validate enterprise solutions forregulated manufacturing industries.Charlton holds a BSc from the University ofGuelph, Ontario, Canada, and has 4 years of

Quality Assurance Lab Experience in the pharmaceuticaland food industries. She has been with PENSA since July2001. She can be contacted at:

Pensa Technology Solutions, Inc., 18 King St. W., Brockville,ON, Canada K6V 3P6, tel: (613) 345-7295, email:[email protected].

Date post:	17-Mar-2020
Category:	Documents
Upload:	others
View:	4 times
Download:	0 times