reprinted from the official journal of ispe …based on the resistance vs. temperature tolerances in...

75
Temperature Measurement Methods SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1 ©Copyright ISPE 2004 Measuring Process Temperature in Small Diameter Lines by Greg Thorp, PE and John Zwak This article presents the advantages and disadvantages of various temperature measurement methods and highlights the use of a Platinum Resistance Thermometer (PRT) to measure the process. Figure 1. Typical surface sensor installation. Preface I n today’s high-performance process sys- tems, measuring the process temperature in small diameter lines down to .25 inch diameter must be understood. The most common applications have processes running in the temperature range of -50°C to +200°C. Measurement uncertainties can easily reach several degrees Celsius over this temperature range due to thermodynamics that can induce conduction effects of the sensor. Other real- world factors, such as the required time re- sponse of the temperature measurement, the ability to replace sensors during process opera- tion, and the ability to Clean-In-Place (CIP) all contribute to the difficulty of this measure- ment. This article looks at several methods that can be used for making these measure- ments including direct immersion (without a thermowell), indirect immersion (with a thermowell), and non-intrusive methods. The temperature sensor assemblies range from a simple clamp-on surface sensor to sensors with elaborately designed thermowells. This article will focus on measuring the process using a Platinum Resistance Thermometer (PRT), due to the performance required for most applica- tions. It also will discuss the advantages and disadvantages of each temperature measure- ment method along with uncertainty estimates based on some common conditions. Introduction Accurate temperature measurement of a fluid flowing in .25 to 4.0 inch diameter lines can be difficult to achieve. While thermocouples, bi- metallic sensors, thermistors, or other devices may be used, they have limitations on perfor- mance that prevent them from meeting the long term accuracy, stability, and repeatability performance available in platinum resistance thermometers (PRTs). Pipes larger than 4 Reprinted from The Official Journal of ISPE PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Upload: others

Post on 11-Mar-2021

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

Measuring Process Temperature inSmall Diameter Linesby Greg Thorp, PE and John Zwak

This articlepresents theadvantages anddisadvantagesof varioustemperaturemeasurementmethods andhighlights theuse of aPlatinumResistanceThermometer(PRT) tomeasure theprocess.

Figure 1. Typical surfacesensor installation.

Preface

In today’s high-performance process sys-tems, measuring the process temperaturein small diameter lines down to .25 inchdiameter must be understood. The most

common applications have processes runningin the temperature range of -50°C to +200°C.Measurement uncertainties can easily reachseveral degrees Celsius over this temperaturerange due to thermodynamics that can induceconduction effects of the sensor. Other real-world factors, such as the required time re-sponse of the temperature measurement, theability to replace sensors during process opera-tion, and the ability to Clean-In-Place (CIP) allcontribute to the difficulty of this measure-ment. This article looks at several methodsthat can be used for making these measure-ments including direct immersion (without athermowell), indirect immersion (with athermowell), and non-intrusive methods. The

temperature sensor assemblies range from asimple clamp-on surface sensor to sensors withelaborately designed thermowells. This articlewill focus on measuring the process using aPlatinum Resistance Thermometer (PRT), dueto the performance required for most applica-tions. It also will discuss the advantages anddisadvantages of each temperature measure-ment method along with uncertainty estimatesbased on some common conditions.

IntroductionAccurate temperature measurement of a fluidflowing in .25 to 4.0 inch diameter lines can bedifficult to achieve. While thermocouples, bi-metallic sensors, thermistors, or other devicesmay be used, they have limitations on perfor-mance that prevent them from meeting thelong term accuracy, stability, and repeatabilityperformance available in platinum resistancethermometers (PRTs). Pipes larger than 4

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 2: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

inches in diameter provide sufficient space for mountingstandard PRT assembly configurations. However, lines from.25 inch to 4 inch diameter require special consideration.Many of the performance advantages of a PRT can be lostthrough improper use or selection of a PRT that is notdesigned for the application.

Industry standards for industrial PRTs, such as ASTME1137 and IEC 60751, are concentrated on cylindricallysheathed, direct immersion style sensors. These documentsprovide no guidance on adapting these thermometer stylesfor use in applications such as the ones described above.Additionally, the performance demonstrated by the sensor ina test laboratory may be completely different than the resultsobtained when used in a production installation.

Measuring temperature in small diameter lines presentssome unique challenges. This article examines several differ-ent methods for measuring temperature in lines down to .25inch diameter, and provides test result for the various meth-ods under some typical conditions.

DiscussionExpectations for PRT SensorsPRT sensors are chosen when process temperature is criticalbecause PRTs offer superior accuracy, stability, and repeat-ability compared to other temperature measuring devices.Many users of PRT sensors have expectations of accuracybased on the Resistance vs. Temperature tolerances in ASTME1137 or IEC 60751, which at 100°C are ±0.3°C for the GradeA or Class A sensors, and ±0.67°C for the Grade B or Class Bsensors. These tolerances apply only to the resistance of thePRT sensor when measured under ideal laboratory condi-tions. In addition, many users request individual PRT cali-bration and transmitter matching which can provide accu-racy to better than ±0.05°C. While these accuracies areachievable in a vast number of process installations, they arenot always achievable in unique installations like smalldiameter lines. In these applications, errors at 100°C couldeasily reach 3°C or larger and can fluctuate greatly depend-ing on ambient conditions.

Temperature MeasurementMethods for Small Diameter LinesThe following four typical methods for measuring the tem-perature of a fluid inside of a line will be discussed in thisarticle:

1. using a surface sensor on the outside of the line

2. installing a non-intrusive sensor in the process line

3. directly immersing a sensor into the fluid flow

Figure 2. Non-intrusive sensor.

Figure 3. Direct immersion sensor installations.

Page 3: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

Figure 4. Indirect immersion sensor installation.

Advantages Disadvantages

Simple to use Slow response time

No line modifications required Highly influenced by ambientenvironment

Installation and replacement are easy Performance variability due tomounting

No need to drain the system to Exposed insulation can bereplace undesirable

Flexibility in location possibilities

No possibility for leaks

No foreign material in process

Low cost

Table A. Surface sensor advantages and disadvantages.

4. installing a thermowell and sensor into the line (referredto as “indirect immersion”)

Surface SensorOne approach to measuring the temperature of the fluidinside the line is to clamp or glue a surface sensor on theoutside of the pipe. Figure 1 shows a sectional drawing of asurface sensor that is attached to a line with an adhesive.This method is one of the simplest to use since the process linedoes not need to be changed by removing a section or addinga port. While adequate performance may be obtained bysimply attaching the sensor to the line, in general, it isrecommended that a thermally conductive adhesive or athermal paste be used between the sensor and line to improvethe heat transfer. By improving the heat transfer, a moreaccurate and faster responding measurement can be made.To further improve the sensor accuracy and minimize theeffects of ambient airflow over the sensor, insulation may beadded over the top of the sensor after installation. Thisdecreases the effects of ambient temperature on the sensor.

Many users find that even with adequate installationprecautions, the surface sensor method is adequate for pro-cess monitoring, but inadequate for control applications dueto measurement errors and slow response times. However,this determination is highly dependent on the particularapplication requirements. Advantages and disadvantages ofthe surface sensor method are listed in Table A.

Non-Intrusive SensorA non-intrusive sensor is typically constructed using a sur-face sensor where the sensing element has been attached toa short section of pipe or tubing that is designed to replace asection of the process line. The sensing element is insulatedand protected by a tube over the section of line. This style ofsensor offers improvements over a standard surface sensorbecause the element mounting and insulation is less variablethan when applied in the field. In addition, the outer sheathprotects the element and insulation from damage and pro-vides for a cleaner installation. Figure 2 shows a sectionalview of a non-intrusive style sensor.

The non-intrusive style sensor solves several of the short-comings of a simple surface sensor and offers improvedaccuracy and response time in a clean package. Advantagesand disadvantages are given in Table B.

Direct Immersion SensorA sensor that is immersed directly into the flow is the typicalsolution that is used on many process lines, large or small,because it provides for accurate measurement and quickresponse time. However, for small diameter lines, the sensorimmersion depth may not be adequate to obtain an accuratetemperature measurement if the sensor is installed perpen-dicular to the flow. A general rule that is used for immersionPRT sensors is that the Minimum Immersion Length (MIL)into the flow should be at least 10 times the sheath diameterplus the length of the sensing element. This immersionlength is required to minimize the stem conduction error, theerror caused by heat transfer between the sensing elementand the ambient conditions at the back of the sensor. For atypical .25 inch diameter sensor with a 1 inch long element,this guideline would require a 3.5 inch immersion into the

Advantages Disadvantages

No immersion into process Response time slower thanimmersion styles

No obstruction of process flow Installations require planning

Element mounting and insulating are Must replace entire pipe section tofactory controlled for consistency replace sensor

Clean external envelope Must drain system to replace sensor

Installation and replacement are easy Calibration can require special baths

Faster response time than simple More expensive than most othersurface sensor PRT sensor options

Table B. Non-intrusive sensor advantages and disadvantages.

“Accuracy and response time can varysignificantly based on the type

of PRT used and the process conditions inwhich it will be used.”

Page 4: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 5. Accuracy comparison graph.

“The user needs to understand the accuracy and response timerequirements in order to choose a measurement method which will

meet the requirements, and avoid unexpected errors dueto misapplication of an otherwise accurate PRT.”

fluid. This may be achievable on lines with a 4 inch or largerInside Diameter (ID), but is not achievable for lines smallerthan this. For lines with an ID smaller than 4 inches, asmaller diameter sensor with a short element length maywork; however, practical limitations on sensor constructionand strength considerations make it difficult to reduce diam-eters to much less than .125 inches. With reduced diameters,the minimum immersion will still need to be approximately1.5 inches.

As an alternate to immersing a sensor perpendicular tothe flow, the sensor may be mounted in the end of a “tee” toallow a longer immersion depth. Figure 3 shows a directimmersion installation in a perpendicular orientation and ina tee with flow parallel to the sensor sheath. While it isconceivable to achieve a proper immersion depth when mount-ing the sensor in a parallel manner, consideration must begiven to other factors such as flow blockage, pressure drop,

and drainability. Advantages and disadvantages of directimmersion sensors are shown in Table C.

Indirect Immersion SensorOne of the most significant disadvantages of the directimmersion sensor is that the system must be shut down anddrained every time a sensor is removed for routine calibrationor replacement. This disadvantage can be eliminated byusing a thermowell, which creates an indirect immersion ofthe sensor. While the addition of the thermowell makesremoval and replacement of the sensor easier, it complicatesthe thermodynamics of the measurement and can increasethe measurement error and slow down the response of thesensor. A sectional view of an elbow with an integralthermowell is shown in Figure 4.

A significant error can occur if the sensor and thermowellare not designed as a system. It is not uncommon for the

Page 5: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

Table C. Direct immersion sensor advantages and disadvantages.

Advantages Disadvantages

Fast response time Installation requires planning

Unaffected by ambient conditions Flow blockage and pressure dropwhen proper immersion is used must be considered

Simple installation Installation could leak

Low cost Stem conduction effects can belarge if proper immersion is not used

Must drain system to replace sensor

Advantages Disadvantages

Sensor is replaceable without Installation requires planningdraining the system

Less affected by ambient conditions Flow blockage and pressure dropthan surface methods must be considered

No possibility for leaks Stem conduction effects can be large

Sensor removal will not introduce More expensive than direct immersioncontaminants into the process methods

Table D. Indirect immersion sensor advantages and disadvantages.

piping designer to specify the installation of an elbow withintegral well where the well is made from standard .375 inchdiameter tubing with a .035 inch wall thickness. While thisstandard size tubing is convenient to use, the resulting wellhas a nominal ID of .305 inch. Typical thermowells specifiedby instrumentation engineers for use with .25 inch diameterPRTs have a nominal ID of .26 inch. A .26 inch ID well, whileless convenient to manufacture, will perform much betterthan a .305 ID well when used with a standard .25 diameterPRT. Further performance improvements can be made byusing a PRT that has been custom designed for use in shortthermowells, or thermowells with an oversized ID. A tipsensitive PRT, one that has a short element length and hasthe element in good thermal contact with the tip of the sensorwhile minimizing the thermal path to the back of the sensor,can improve accuracy and response time of the measurement.Smaller diameter thermowells with correspondingly smallerPRTs are available as well, when the application demands it.

The advantages and disadvantages of the indirect immer-sion method are listed in Table D.

Analysis of Measurement ErrorsIt is beyond the scope of this article to discuss the mathemati-cal models that describe the heat transfer relationships for allthe different styles of sensors. Suffice to say that when a highquality PRT is used to measure temperature, in most instal-lations and under steady state conditions, thermal conduc-tion effects are the dominant source of error. This error isdirectly related to the magnitude of the difference in tem-perature between the fluid being measured and the ambientsurroundings, this difference is referred to as “Delta T” (∆T).It can be shown that under steady state conditions theconduction error may be represented as a percent of ∆T. Avery simplistic interpretation is to view every installation ofa PRT as having a thermal profile that goes from “near fluid”temperature to ambient environment temperature. The goalfor an accurate temperature measurement is to have the PRTsensing element located in the “near fluid” portion of thisprofile. The best way to accomplish this is to thermally couplethe PRT element to the process fluid, and thermally isolatethe PRT element from the ambient environment.

ExperimentationAccuracy TestingAccuracy testing was performed in a laboratory controlled“sample process” on several of the different types of sensors

PRT sensor type Error Error(% of ∆T) (% of ∆T)

No ambient With ambientairflow airflow

Surface Sensor - clamped on with:No thermal compound, no insulation 9.7% 26.6%No thermal compound, insulated 7.0% 8.0%With thermal compound, no insulation 4.7% 11.1%With thermal compound, insulated 1.8% 2.0%

Non-Intrusive 0.7% 2.4%

Direct immersion (.125 diameter x 1.0 long) 0.2% 0.2%

Indirect immersion (.25 dia PRT):.305 ID well - standard PRT 3.7% 6.3%.305 ID well - Tip sensitive PRT 0.4% 0.8%.260 ID well - standard PRT 0.3% 0.6%.260 ID well - Tip sensitive PRT 0.3% 0.0%

Table E. Accuracy test results for 4 different sensors on ½ inchdiameter line. Water at 50°C (∆T = 28°C), 3 feet per second.

previously described in this article. The “sample process”that was used to test these sensors was hot water flowing atapproximately three feet per second through a .5 inch outsidediameter stainless steel tube with a 0.065 wall. A briefdescription of the sensors that were tested are as follows:

Surface sensor - a typical clamp-on style surface sensor wastested as installed with and without thermal compound atthe sensor to line interface, and with and without insulationover the sensor.

Non-intrusive sensor - a sensor of construction similar to thatshown in Figure 2 with a .5 inch diameter process line witha 0.050 wall.

Direct immersion sensor - a .125 inch diameter by 1.0 inchlong sensor with a sanitary cap process connection installedperpendicular to the flow.

Indirect immersion sensor - a process line elbow with an inte-gral thermowell similar to that shown in Figure 4. The wellwas approximately 2 inches long by .375 inch OD with eithera .305 or .260 ID and was tested with both a standard .25 inchdiameter PRT and a tip sensitive .25 inch diameter PRT.

All of the PRTs used for this testing were calibrated atmultiple temperature points using a method typical forindustrial PRT calibration, the uncertainty of the calibrationwas estimated not to exceed 0.025°C. This calibration was

Page 6: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

6 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

PRT sensor type 63.2% Response

Clamp On SurfaceNo thermal compound, no insulation 38 secondsNo thermal compound, insulated 55 secondsThermal compound, no insulation 17 secondsThermal compound, insulated 20 seconds

Non-Intrusive 11 seconds

Direct immersion (.125 diameter x 1.0 long) <5 seconds

Indirect immersion (.25 dia PRT):.305 ID well - standard PRT 91 seconds.305 ID well - Tip sensitive PRT 40 seconds.260 ID well - standard PRT 30 seconds.260 ID well - Tip sensitive PRT 22 seconds

Table F. Response Time Test Results for 4 different sensors on ½inch diameter line. Step change in water from 22°C to 50°C, 3feet per second.

Figure 6. Response time comparison graph.

performed so that the individual resistance vs. temperaturecharacteristics for each sensor could be used to accuratelycalculate the temperature and determine the measurementerror. The sensors used for monitoring the water supply andambient air temperatures were a secondary standard gradePRT that was matched to a precision digital thermometer,the overall accuracy of these monitoring systems is estimatedto be less than 0.035°C.

Two variations of the test were performed. The firstvariation was at a fluid temperature of approximately 50°Cwith no ambient airflow over the portion of the sensor outsidethe process. The second condition was similar to the firstexcept a 4 inch diameter fan was placed 12 inches from thesensors to circulate ambient air over the portion of the sensor

outside the process. This was done to determine the sensitiv-ity of the measurement method to ambient conditions. Theresults of the test are given in Table E and shown graphicallyin Figure 5. To account for variation in fluid temperatures,the errors are presented as a percent of ∆T. Presenting theresults as a percent of ∆T not only normalizes the data, butalso allows the data to be used to estimate expected errorsunder different temperatures. For example, a sensor thatexhibited an error of 0.7% of ∆T, if used in a 121°C steriliza-tion process (with similar heat transfer characteristics) wouldhave an estimated error of 0.7°C (0.7% of the 100°C ∆Tbetween the process temperature and ambient surroundingtemperature). Therefore, the accuracy recorded can be usedto approximate the actual uncertainty in a 121°C steriliza-tion process.

Response Time TestingA test was conducted to determine the relative response timeof the various PRT measurement methods. The test wasconducted by pumping hot water through the lines, whichwere initially at room temperature, and determining howlong each measurement method took to reach 63.2 percent ofthe step change in temperature. Since all methods weretested using identical flow conditions, a direct comparisoncan be made between methods. It is important to note thatactual installation conditions will significantly affect thisresult so this test was meant to give relative informationonly. The results are given in Table F and shown graphicallyin Figure 6.

Page 7: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Temperature Measurement Methods

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 7©Copyright ISPE 2004

ConclusionTo achieve accurate temperature measurement in lines from.25 to 4 inches in diameter requires special consideration.Standard PRT sensors do not perform the same in small lineinstallations as they do in calibration baths in laboratoriesprimarily due to thermal conduction effects and ambientenvironment influences. Accuracy and response time canvary significantly based on the type of PRT used and theprocess conditions in which it will be used. The user needs tounderstand the accuracy and response time requirements inorder to choose a measurement method which will meet therequirements, and avoid unexpected errors due to misappli-cation of an otherwise accurate PRT. Direct immersion PRTsshould be considered where the highest accuracy is requiredfor control of the process temperature. Non-intrusive orsurface mount PRTs may be used where best accuracy is notrequired, such as process monitoring. Additional require-ments such as the convenience of clamping a surface sensoron to the outside of a line, or the ability to remove a sensorwithout draining the system also will impact the final deci-sion. With the proper choices, an accurate, stable, and repeat-able measurement can be achieved.

About the AuthorsGreg Thorp’s responsibilities at Burns En-gineering include the configuration, techni-cal support, sales, and marketing of tem-perature sensors. He also focuses on manag-ing projects at Burns Engineering and hasspecialized on temperature measurementsin small diameter lines. He has more than 10years of experience in the design, inspection,

and evaluation of above ground storage tanks, pressurevessels, and piping systems. He has received a BS and MSEngineering degrees from the University of Minnesota. He isa member of ISPE and is an ASME-BPE committee member.He can be contacted by e-mail: [email protected].

John Zwak received his BS in aerospaceengineering and mechanics from the Univer-sity of Minnesota. He joined the Tempera-ture Division of Rosemount Inc. in 1984where he held various levels of responsibilityin the design, development, manufacture,and testing of temperature sensors for aero-space, industrial, and nuclear applications.

He joined Burns Engineering in 2000 and is currently theDesign Engineering Supervisor where his primary responsi-bilities include development and testing of new platinumresistance thermometers and thermocouples for high perfor-mance industrial and metrology applications. He is an activemember of ASTM International committee E20 on Tempera-ture Measurement. He can be contacted by tel: 1-952/935-4400 or by email: [email protected].

Burns Engineering, 10201 Bren Road East, Minnetonka,MN 55343.

Page 8: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

Biotech CIP Cycle Developmentby Timothy Howard and Matt Wiencek

This articledemonstrateshow theprinciples ofISPE's C&QBaseline® Guidecan be appliedto a cycledevelopmentprogram at amodern biotechmanufacturingfacility.

Figure 1. Overview ofCD testing program andcleaning validationprerequisites.

Introduction

CIP Cycle Development (CD) is a sys-tematic approach of setting CIP sys-tems to work in a manner that pro-motes rapid and successful execution

of Cleaning Validation (CV) activities. TheCycle Development program is an integral partof the overall commissioning of a facility and itsprocesses. It is not enough to merely confirmtubing, valves, pumps, controllers, and instru-mentation are installed as designed, function-ing as specified, and within acceptable toler-ances. The act of running a cleaning cycle, andconfirming the software and associated controlelements respond as expected (or functionaltesting) does not constitute Cycle Development.All of these activities are elements that musttake place prior to Cycle Development, but donot provide for development of a cleaning cycle.Cycle Development employs CIP skids thathave been commissioned, and through a three

step process, optimizes the cycle parametersfor each cleaning circuit in the plant. A formalCycle Development program will lead to robustCIP cleaning cycles, which in turn minimizescleaning validation deviations, retests, andlengthy investigations of failures. The purposeof this article is to demonstrate how the prin-ciples of the ISPE Commissioning and Qualifi-cation (C&Q) Baseline® Guide can be applied toa Cycle Development program at a modernbiotech manufacturing facility.1

BackgroundCycle Development (CD) typically occurs afterCIP and support systems are mechanicallycomplete and prior to Cleaning Validation (CV)execution. CD is one subset of activities withinthe overall commissioning effort. The ISPEBaseline® Guide on Commissioning and Quali-fication (C&Q) can be applied to plan and ex-ecute an organized and efficient CD program.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 9: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

This article will examine the planning of a CD program for amodern biotechnology manufacturing facility and presenttwo case studies for execution of the plan. The case studieswill examine development of a process vessel cleaning circuitand a Tangential Flow Filter (TFF) skid cleaning circuit.

For the purposes of this case study, the CIP system isdeemed to be a “direct impact” system as defined in the C&QGuide.2 The CIP system includes the CIP skid, supply andreturn piping, process equipment, and the instrumentationused to monitor critical parameters of the cycle. Direct Im-pact systems require full commissioning prior to or inte-grated with qualification activities. The Baseline® Guidedefines commissioning as:

A well planned, documented, and managed engineeringapproach to the start-up and turnover of facilities,systems, and equipment to the End-User that results insafe and functional environment that meets establisheddesign requirements and stakeholder expectations.3

The Commissioning Program is defined by a formal Commis-sioning Plan which lists a set of deliverables. The C&Q Guidedefines these as follows:4

• Commissioning Schedule• Budget, Pre-Delivery Inspection Plan, and Report• Factory Acceptance Plan and Report• On-Site Inspection Plan and Report• Functional Test Plan and Report• Commissioning Plan Summary Report.

The C&Q Guide describes the activities contained in theFunctional Test Plan as follows:5

• setting equipment to work• regulation and adjustment

• performance testing

CIP Cycle Development activities fall into the category ofPerformance Testing, and therefore should be defined withinthe Functional Test Plan of the Commissioning Program. CDrequires a fully functional, integrated control system/physi-cal installation. CD may result in changes to either thephysical configuration of a system or to the cleaning sequenceof operations. Therefore, it should occur after the functionaltesting phase of commissioning, but prior to formal opera-tional qualification. Performing at least some CD beforeoperational qualification (although perhaps not the soiledphase) will reduce the necessity for strict change control andthe number of qualification deviations that may need to beprocessed as a result of changes to operational sequences orphysical configuration – Figure 1.

Developing the PlanThe first step in developing a CD test plan is to define thefunctional boundaries. The functional boundaries can easilybe defined as the individual CIP cleaning circuits. Today’sbiotechnology facilities have multiple CIP skids with eachone typically dedicated to a manufacturing department orsubset of similar equipment. In the example discussed below,one CIP skid services each of the following areas: Buffer Prep,Media Prep, Fermentation, and Purification. Each of theseCIP skids will serve multiple cleaning circuits in its respec-tive area. The exact number of circuits should be defined in aFunctional Specification (FS). The number of cleaning cir-cuits is based on the amount and type of unit operations in theplant. A CIP Cycle Development Test Plan for the Buffer Preparea is defined in Table A. Similar plans should be generatedfor the Media Prep, Fermentation, and Purification areas.

The process P&IDs will define the major equipment sys-tems (column 1 of Table A). The FS will define the CIP circuittype and the unique circuit name (column 2 and 3 of Table A).Each circuit can be designated as “unique” or “clone” (column4 of Table A). This designation is a necessary element of thetest plan for accurate resource planning purposes. For ex-ample, two identical vessels located next to one another in themanufacturing area, and served by the same CIP Skid, maybe considered the same for CD purposes. The 500L BufferPrep Vessels in Table A meet these criteria. The “unique” or“clone” designation can only be made after a careful analysisof the equipment, size, location, configuration, soil type, andcleaning chemistry. The “clone” circuits will require less CDfield work because programmable recipe parameters will becopied from the unique circuit. It is important to note that acircuit considered to be a clone during CD may or may not beconsidered a clone when performing cleaning validation stud-ies.

Generating a ScheduleA facility test matrix can then be developed to aid in thecalculation of manpower requirements - Table B. This isnecessary for generating a realistic Commissioning Scheduleas required by the C&Q Guide.

Equipment CIP Circuit Circuit Unique/Name Clone

500L Buffer Prep Vessel #1 Tank B-C01 UniqueInlet Line B-C02 Unique

Outlet Line B-C03 Unique500L Buffer Prep Vessel #2 Tank B-C04 Clone(Vessel identical to #1) Inlet Line B-C05 Clone

Outlet Line B-C06 Clone2000L Buffer Prep Vessel #3 Tank B-C07 Unique

Inlet Line B-C08 UniqueOutlet Line B-C09 Unique

1000L Buffer Hold Vessel #4 Tank B-C10 UniqueOutlet Line B-C11 Unique

2000L Buffer Hold Vessel #5 Tank B-C12 UniqueOutlet Line B-C13 Unique

2000L Buffer Hold Vessel #6 Tank B-C14 Clone(Vessel identical to #5) Outlet Line B-C15 Clone3000L Buffer Hold Vessel #7 Tank B-C16 Unique

Outlet Line B-C17 Unique

Table A. Functional test plan - buffer prep.

Page 10: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

CD consists of three distinct phases: water cycle testing,chemical (or cleaning agent) cycle testing, and soiled equip-ment testing. This incremental approach assures that chemi-cals are introduced into the systems only when it is safe to doso. The effectiveness of the pre-qualification soiled test runscan be measured by collecting rinse samples. The rinsesample data can be used to establish a cleaning data bankprior to starting the Cleaning Validation program. The man-power estimates per task (shown in man-days in Table B)represent the amount of time one member of the CD team willspend in the field running the CIP skid and process equip-ment. Each facility should use estimates that are particularto their site. The manpower requirements shown in Table Bare estimates. The Buffer Department estimates are shownin full detail. Again, similar tables should be developed forthe Media Prep, Fermentation and Purification Departments.We will assume that the overall Commissioning Plan desig-nates a three month window in which the CD must becompleted. Therefore, the following staffing estimates can bedeveloped:

205 work-days / 20 work-days/month = 10 work-months

10 work-months / 3 months permitted by schedule ~4 team members full time to satisfy the schedule.

The initial reaction to the required field execution time forCycle Development is typically a surprise to those not famil-iar with the CD test procedures. The amount of time neces-sary for this effort often leads to multiple shift or around the

clock testing. When scheduling the CD activities, it is criticalto understand resource limitations that may negatively im-pact the effort. These limitations may include the capacity ofwater systems to meet demand and availability of QC labora-tory coverage for testing. Most water systems are not de-signed to support continuous use of all CIP skids around theclock. Similarly, during the startup phase of a facility, QClaboratories rarely can support sustained around the clocktesting of this nature. Equally important in scheduling theCD activities is to assess the impact of other plant or processcommissioning. It is impossible to conduct cycle developmenton a media prep vessel while the batch software for the vesselis being commissioned. With a good understanding of theselimitations, and the magnitude of the effort required, anaccurate CD schedule can be integrated with the entirefacility commissioning schedule.

CIP Cycle Development Test ProceduresThe CD test procedure should define the amount of “regula-tion, adjustment, and performance testing” as required bythe C&Q Guide. These procedures should be written tocomply with the requirements of the FS for each unit opera-tion. Their purpose is to control start-up/CD activities onlyand are not to be considered GMP documents. The FS mustdetail the recipe parameters for each piece of equipment to becleaned. By using the incremental approach of water batch,chemical batch, and soiled batch testing, the CD team canefficiently and safely define these parameters for the Clean-ing Validation effort.

Circuit Name CIP Circuit Type Unique/Clone Work-days per Work-days per Work-days per Total Work-days ofWater Test Run Chemical Test Run Soiled Test Run CD Field Work

B-C05 Inlet Line Clone 0.5 0.25 0.25 1B-C02 Inlet Line Unique 1 0.5 0.5 2B-C09 Inlet Line Unique 1 0.5 0.5 2B-C07 Outlet Line Clone 0.5 0.25 0.25 1B-C15 Outlet Line Clone 0.5 0.25 0.25 1B-C03 Outlet Line Unique 1 0.5 0.5 2B-C10 Outlet Line Unique 1 0.5 0.5 2B-C12 Outlet Line Unique 1 0.5 0.5 2B-C13 Outlet Line Unique 1 0.5 0.5 2B-C17 Outlet Line Unique 1 0.5 0.5 2B-C04 Tank Clone 1.5 0.5 0.5 2.5B-C14 Tank Clone 1.5 0.5 0.5 2.5B-C01 Tank Unique 3 1 1 5B-C08 Tank Unique 3 1 1 5B-C11 Tank Unique 3 1 1 5B-C12 Tank Unique 3 1 1 5B-C16 Tank Unique 3 1 1 5

Total Man-days of CD Field Work in Buffer Prep 47Total Man-days of CD Field Work in Fermentation 55Total Man-days of CD Field Work in Media Prep 59Total Man-days of CD Field Work in Purification 44Total Man-days for CIP Cycle Development 205

Table B. Biotech facility CIP cycle development work estimate.

Page 11: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Line CleaningDepartment Contact Time(minutes) Temperature Supply Conductivity Cleaning Action

Buffer 3 minimum Ambient rinse WFI rinse 3-5x holdup volume turnoverAmbient wash WFI wash at 5ft/sec velocity minimum

Media 3 minimum Ambient rinse WFI rinse60oC wash 0.6 mS

Fermentation 5 minimum Ambient rinse WFI rinse60oC wash 0.6 mS

Purification 5 minimum Ambient rinse WFI rinse60oC wash 0.3 mS

Vessel CleaningDepartment Contact Time(minutes) Temperature Supply Conductivity Cleaning Action

Buffer 3 minimum Ambient rinse WFI rinse Achieve FAT sprayball ratingsAmbient wash WFI wash for flow and pressure.

Media 3 minimum Ambient rinse WFI rinse Visual confirmation that60oC wash 0.6 mS surfaces are residue free.

Fermentation 5 minimum Ambient rinse WFI rinse60oC wash 0.6 mS

Purification 5 minimum Ambient rinse WFI rinse60oC wash 0.3 mS

TFF CleaningDepartment Contact Time(minutes) Temperature SupplyConductivity Cleaning Action

Rinse : MembraneRinse : 5 minimum Rinse : Ambient Rinse : WFI rinse pressure setpoint(s) and

tangential velocityAll Departments Soak : 30 minimum Soak : Hold time

Soak and Recirculate Wash : Soak and Recirculate Wash : Recirculate wash : MembraneRecirculate Wash : 10 minimum Membrane dependent Membrane dependent pressure setpoint(s) and

tangential velocity

Table C. Cleaning acceptance criteria.

The CD team must understand acceptance criteria foreach cleaning circuit before proceeding with testing. Thefundamental acceptance criteria should include:6

• Contact Time: defines the amount of time, usually inminutes, that the rinse and cleaning solution are in con-tact with soiled surfaces.

• Cleaning Temperature: the temperature of the solutionsupplied to the equipment from the skid. The tempera-tures required are unique to each cleaning circuit. Typi-cally, rinses are conducted with ambient WFI.

• Solution Conductivity: the conductivity of the solutiondelivered to the equipment. The required value is a func-tion of the detergent utilized.

• Cleaning Requirement: unique for each unit operation,but typically can be defined as the amount of fluid turbu-lence, fluid impingement, and/or soak time.

• Contaminent Solubility: it is important to ensure thatthe contaminent is soluble in the rinse solvent which issubsequently tested to demonstrate purity.

Table C defines the water and chemical batch cleaningacceptance criteria for three unit operations utilized duringCD. The acceptance criteria for soiled equipment should bedefined by the Validation Master Plan.

Now that the cleaning acceptance criteria are defined, theCD team must go about “setting the equipment to work” tomeet the requirements. The regulation and adjustment isaccomplished by configuring the programmable recipe pa-rameters of the software. The recipe parameter names can beextracted from the Functional Specifications for each unitoperation. The CIP skids also have their own set of param-eters. The number of configurable parameters for each unitoperation is on the order of 20-30. If Table A is completed forall four CIP skids, there would be six unique unit operationsand a total of approximately 70 unique circuits. Therefore,the number of unique parameters that need to be tracked bythe CD team is substantial, on the order of 2,000 uniquenumeric values. Utilizing a structured approach as defined inthe C&Q Guide becomes the only way to manage the effort.

Tracking CIP Configurable ParametersDuring Cycle Development

The objective during CD is to identify and programconfigurable parameters for each CIP circuit that will beutilized during the formal CV testing. These parameters will

Page 12: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

Unique Parameters Test ResultsCircuit # Water/ P-1(gpm) P-2(sec) P-3(min) P-4(gal) P-5(mS) Rinse Swab Date

Chemical/ Sample (uS) Sample (ppm)Soil Test

B-01 Water NA NA NAB-01 Chemical NAB-01 SoilB-02 Water NA NA NAB-02 Chemical NAB-02 Soil

Table D. Buffer department unique configurable parameter and test results database.

provide a high degree of assurance that the cleaning valida-tion effort will be consistent and successful. Employing adatabase to document parameter values and test results isnecessary.

The first step is to identify parameters that are unique toa CIP circuit from those that may be commonly used withina manufacturing department or grouping of equipment. Forexample, the wash volume required, air blow time, andgravity drain time are generally unique to each circuit. Thewash conductivity, conductivity deviation allowance, andfinal rinse conductivity are generally common to multiplecircuits. By separating the unique and common parameterswithin the database, a consistent approach to CV can bedeveloped throughout the manufacturing facility. The data-base may take the following form and assumes that once thecommon values are defined, only the unique values will betracked along with the test results. The Buffer Departmentdatabase format is defined in Table D. Similar tables shouldbe developed for each department.

Executing the Plan -Case Study 1 - Process Vessel

It should be noted, that the following “Case Studies” aregeneral and should not be applied as a “rule” to all manufac-turing situations. Each manufacturing site needs to apply theCD principals in a manner that is appropriate to the process,the equipment being cleaned, and the biological soil in ques-tion.

Phase 1 - Water BatchThe objective of the water cycle is to achieve rated flow andpressure for the rinse and wash sprayball(s) cycle timeswhich ensures proper coverage. This may require the place-ment of a pressure indicator near the sprayball port on thetank. The flow measurement can be referenced from the CIPskid. Sprayball design parameters and the results of FactoryAcceptance Test (FAT) sprayball testing should be referencedwhen determining the required flow rates and pressure.Tanks may have multiple inlet pathways that are within thecleaning boundary of the CIP circuit. The CD team will needto optimize the valve sequencing patterns to meet the clean-ing acceptance criteria for lines as part of the vessel cycle.

CIP skid tanks and the overall CIP circuit have fixed hold-up volumes. The hold-up volume of the CIP rinse and washtanks should be large enough to fill the circuit and provide

enough extra volume to make-up for losses to drain duringthe rinse and wash steps. The typical CIP cycle consists of aseries of rinse-wash steps with intermediate gravity drainsand air blows to remove spent solution. The air blow time andgravity drain time are determined in the field by collectingreal time data. In some cases, a CIP Return (CIPR) pump isutilized to deliver cleaning solution back to the CIP skid drainor wash tank. The pump must be turned on at the correct timeto prevent excessive pooling in the vessel, but not beforeproper priming has been achieved. Likewise, the CIPR pumpmust be shut off after removing solution from the circuit, butprior to running dry for an excessive amount of time.

Water Batch Test ProcedureDetermine the following recipe parameters: rinse time, washtime, CIP supply flow rate and pressure, CIP supply rinse andwash volumes, gravity drain and air blow time, vessel inletpathway valve cycling time, valve to drain cycling time, CIPRpump start time delay, and CIPR pump stop time delay. Theapproximate total cleaning time for each circuit also may becalculated at this point in CD.

Phase 2 - Chemical BatchChemical testing can begin once water batch testing demon-strates the CIP skid and process equipment function as anintegrated CIP circuit. Chemicals are added to solution at theCIP skid, typically into the wash tank during recirculation ofthe solution. The initial “bulk” charge of chemical will drivethe solution conductivity to roughly 90% of the requiredvalue. Incremental additions are then made bringing thesolution to the required conductivity. The wash solution isthen recirculated and heated in the CIP skid to achievehomogeneity with respect to conductivity and temperature.Once temperature and conductivity set points are achieved,the solution is delivered to the vessel and returned to the skid.The hold-up mass of the equipment will “rob” the system ofenergy and the initial return temperature of solution will beless than required. Therefore, the wash timer will not startuntil the solution and equipment are “up to temperature.” Ifvalves are cycling to drain and creating losses, the hold-upvolume of the wash tank must be large enough to compensatefor the losses. Therefore, wash volumes can be greater thanrinse volumes. Once the washes and intermediate rinses arecomplete, the final rinse should bring the return conductivitymeasurement to within specification, typically that of WFI.

Page 13: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

6 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

The purpose of the chemical test is to verify that cleaningagents are not trapped by “dead legs,” all surfaces are “freedraining,” and to test the chemical delivery system on the CIPskid.

Chemical Batch Test ProcedureDetermine the following recipe parameters: chemical pumpbulk delivery time, chemical pump incremental deliverytime, tuning parameters for wash heat-up, wash volumebased on additional losses to drain, final rinse time.

Phase 3 - Soiled Batch TestingUnfortunately, after all the previous cycle development, theCD team will not know whether the CIP cycle will satisfy therequirements of cleaning validation until tested on biologicsoil. The difficulty of cleaning the soils is a function of thesoil’s fouling propensity, the hold up volume of the circuit,and its mechanical configuration. The soiled test procedureshould include instructions on how to obtain rinse and swabsamples. The rinse sample can be obtained by drawing froma sample port at the CIP skid on the CIPR line. Otherlocations can be determined as optimal. CIP skids often arenot designed with these ports in mind so the CD team willneed to figure out how to have a sample apparatus fabricatedand draw the sample in a way that does not compromise thesample quality. The sample should be measured for con-ductivity and Total Organic Carbon (TOC). TOC limits areestablished by the Cleaning Validation group and reflect themaximum acceptable level in the final rinse. Difficult to cleanareas of the circuit should be swabbed to measure the effec-tiveness of the cycle. Once the analytical results are obtained,the CD team must decide if the programmed recipe param-eters need to be adjusted. If the circuit is still “dirty,” the CDteam can optimize the recipe parameters for better cleaning.If the circuit is “clean” the recipe parameters may be adjustedto decrease the overall CIP cycle time and thus increase plantthroughput.

Soiled Batch Test Procedure1. Run CIP cycle on soiled equipment.2. Obtain rinse and swab samples.3. Optimize recipe parameters to achieve required cleaning

in minimal amount of time.4. Determine if mechanical or software changes are required

to achieve cleaning objectives and implement before thestart of CV.

Executing the Plan -Case Study 2 - TFF Skid

Phase 1 - Water BatchCleaning circuits for TFF skids are configured in a variety ofways. In this case study, assume that cleaning solutions aredelivered by the CIP skid to the process hold vessel utilizedby the TFF unit during normal operations. All solutions arefirst charged to the vessel incrementally before a CIP cyclestep begins. Rinse solutions are pumped from the hold vessel,by the process pump, through the TFF skid directly to a local

drain. Wash solutions are recirculated through the mem-branes back to the hold tank before being sent to the samelocal drain. Therefore, the hold up volume of the process tankwill determine the length of each wash and rinse step.Multiple rinse/wash steps may be needed to achieve desiredcontact time depending on the fluid throughput.

The cleaning action for a TFF unit is defined by thetangential flow velocity achieved across the membrane, therequired trans-membrane pressure drop, and membranesoak time. The tangential velocity will “sweep” away biologicsoils. The trans-membrane pressure differential will createflux flow through the membrane. Some membranes aresensitive to extreme temperatures and therefore will onlyaccommodate ambient rinses and washes. The required tan-gential velocity, trans-membrane pressure drop and cleaningsolution temperature should be referenced from manufactur-ers’ recommendations or plant operating experience.

Water Batch Cycle Test Procedure1. Determine required wash and rinse volumes.2. Determine the TFF pump speed necessary to achieve

tangential flow velocity.3. Tune the flow and pressure drop control loops.4. Adjust valve pathways to direct solutions to recycle and

drain.

Phase 2 - Chemical BatchThe membrane material and biologic soil will determine thetype and concentration of cleaning solutions required. In thisexample, the CIP skid will batch wash solutions to the holdvessel without CIPR flow. Therefore, the CIP skid recipeparameters must be adjusted accordingly. The chemicaldelivery algorithm at the CIP wash tank will be the samehowever. Soaking the membrane in wash solution is accom-plished by closing isolation valves on the inlet and outletports of the membrane during the wash cycle.

Chemical Batch Cycle Test Procedure1. Determine the required wash solution conductivity and

temperature.2. Adjust the valve pathways for a recirculation, drain, and

soak step.3. Ensure enough wash solution is delivered to hold tank.4. Adjust CIP skid operating parameters for no CIPR flow.

Phase 3 - Soiled Batch TestingThe CIPR conductivity sensor cannot be utilized to determineif the final rinse has flushed all soil and cleaning solutionfrom the equipment. The CD team must rely on rinse samplesto confirm the rinse time and volume utilized by the TFF unitfor a final rinse are adequate. Surface interactions at mem-brane fluid interfaces are a function of chemical absorptionand typically cannot be cleaned purely by variation of thecross flow velocity. The main cleaning action for bio productsis achieved by surface active agents as is the case for allsorptive processes. The circuit differs from the tank cleaningin that it will be validated based upon final rinse time and not

Page 14: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

CIP Cycle Development

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 7©Copyright ISPE 2004

CIPR conductivity. A local conductivity sensor also might beutilized for this situation.

Soiled Batch Test Procedure1. Run CIP cycle on soiled equipment.2. Determine final rinse time and volume required by sam-

pling.3. Optimize recipe parameters to achieve required cleaning

in minimal amount of time.4. Determine if mechanical or software changes are required

to achieve cleaning objectives and implement before thestart of CV.

This series of procedural steps should be repeated for everytype of unit operation concentrating first on the “unique”circuits. Cloned circuits may proceed directly to soiled test-ing. Each unit operation will have a unique set of recipeparameters to track. Typically, line and tank cleaning are themost straightforward while TFF skid, chromatography col-umns, and centrifuges present more difficult challenges. Inevery case; however, the Functional Specification will definethe configurable parameters that must be incorporated intothe CIP Cycle Development Test procedures for the CD team.

SummaryThis approach to CIP Cycle Development requires implemen-tation of the C&Q Guide, developing a detailed plan based onequipment functional specifications, accurately schedulingthe activities, and implementing three phases of testing.When successfully implemented, CIP cycle development willpromote a very successful cleaning validation effort, andultimately minimize production losses related to insufficientcleaning cycles.

References1. ISPE Baseline® Pharmaceutical Engineering Guide, Com-

missioning and Qualification, Volume 5, 2001.2. Ibid, p.30.3. Ibid, p. 19.4. Ibid, p.53.5. Ibid, p.57.

6. Brunkow, R., Delucis, D., Haft, S., Hyde, J., Lindsay, J.,McEntire, J., Murphy, R., Myers, J., Nichols, K., Terranova,B., Voss, J., and White, E., Cleaning and Cleaning Valida-tion: A Biotechnology Perspective, PDA, Bethesda, MD,1996.

About the AuthorsTimothy P. Howard, PE has more thaneight years of experience in pharmaceuticaland biotechnology validation and commis-sioning. He is a former Navy submarineofficer, previously licensed as a senior reac-tor operator for a commercial nuclear powerplant, and is a registered professional engi-neer in North Carolina. He has a degree in

mechanical engineering from North Carolina State Univer-sity. Howard is a board member of the ISPE Carolina SouthAtlantic Chapter and serves on the ISPE North AmericanContinuing Education Committee. Howard co-authored thisarticle while employed with Commissioning Agents, Inc. Heis currently employed with Eli Lilly & Co. He can be contactedby e-mail: [email protected].

Eli Lilly & Co., 9700 Innovation Loop, Manassas, VA20110.

Matt Wiencek holds a BS in chemical engi-neering from Cleveland State University anda Bachelor of Environmental Design fromMiami University. He has been a ValidationEngineer and Project Manager with Com-missioning Agents, Inc. since 2000, and hasalso held positions as a process design engi-neer and architectural consultant. Wiencek

has accomplished project work in the areas of plant commis-sioning of biotechnology and bulk pharmaceutical chemicalfacilities. He is currently a Board Member of the ISPE GreatLakes Region Chapter. He can be contacted by tel: 1-440/653-1577 or by e-mail: matt.wiencek @cagents.com.

Commissioning Agents, Inc., 1515 N. Girls School Rd.,Indianapolis, IN 46214.

Page 15: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

Planning and Operational Aspects ofRobotics Laboratoriesby Paul Leonard and Lou Angelus

This articlepresents drugdiscoveryresearch trendsand identifiesthe impact onlaboratoryfacility design,construction,and operation.

Figure 1. Traditionalchemistry laboratoryfloor plan.

Understanding Research DriversBehind Robotics Laboratory

Design

Under constant pressure to dramati-cally improve R&D productivity in aclimate of rising R&D costs, shorterproduct lifecycles and limited sales

growth, biotechnology researchers are turningto robotic processes and equipment to not onlyincrease the quantity of compounds discov-ered, but also to enhance the quality of theleads created via the discovery process.

The profiling and screening portion of drugdiscovery is currently one of the best candi-dates to utilize automated research processes.This type of research typically occurs very earlyin the discovery process and its purpose is toidentify compounds with specific biological ac-tivities required for potential use against atargeted disease. This high throughput type ofcompound profiling and screening process, itsassociated equipment, and the effects on the

design of laboratory facilities are the topics ofthis article.

Profiling typically tests multiple biologicalactivities within a prescribed class of com-pounds. This process can discover multiple ben-efits or detriments associated with the targeteddisease. Profiling may also uncover additionalbiological activities that benefit multiple sepa-rate drug development projects. Screening testsone specific biological activity in a large num-ber of compounds; therefore, testing multiplecompounds for that one specific activity or formultiple activities.

Automation has led to faster and more cost-effective profiling and screening with morereliable results. The general benefits of auto-mation are uniformity and the minimization ofvariables that leads to higher data quality, andtherefore, allows the researcher more time tointerpret the data. With known uniform inputprocedures, less analysis is typically requiredto be performed because the researcher can

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 16: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

concentrate on the particular reaction and not reactions frominput error. This equipment evolution also is having anadditional benefit by reducing the usage of solvents, com-pounds, and enzyme materials in the discovery process, thusreducing waste streams.

Even as little as 10 years ago, research for new compoundsand the testing for interactions was typically conducted by aresearch chemist in a fume hood intensive lab. With today’scombinatorial chemistry techniques, the same chemist hasincreased the speed of discovering and testing of compoundsby as much as tenfold in the same fume hood intensive labenvironment. The high throughput screening process is fur-ther accelerating the testing of compounds beyond the limitsof combinatorial chemistry and has produced as much as ahundredfold increase in the amount of compounds that can bescreened. The design of robotic laboratories is required torecognize how this trend will continue and how to anticipatethe flexibility to allow for alternative equipment arrange-ments.

This mechanization of the drug discovery process willcontinue as the pressures of the industry continue to push forincreased efficiency of discovery. Equipment technology iscontinuing to develop in a form that further minimizes theuse of materials via micro-fluidics. Micro-fluidics and “Lab-on-a-Chip” technologies are evolutions of the current “welland plate” high throughput screening technology. This tech-nology uses much less reagent than would be required for theequivalent plate-based assay. For example, for cell-basedassays, it is possible that a 100 times fewer cells per datapoint are required compared to their equivalent plate-basedassays. For scientists constrained by limited cell availability,e.g., those working with primary cell lines, this technologyoffers a way to perform more assays than would otherwise bepossible.1 With the continued drive to reduce R&D costs, it

has been predicted that successful drug discovery efforts inthe not so near future will depend on the further mechaniza-tion of the discovery process, greater reliance on affiliationsbetween companies, more effective management of the database of compounds that is being created, and continuedmovement to computer generated molecular libraries,bioinformatics, and virtual clinical trials. As today’s typical20-25 percent research completed via automated methodscontinues to grow, the role of chemistry research and associ-ated facilities will need to change dramatically.2,3

Laboratory Facility Design ResponsesThe research trends described previously are driving newplanning criteria and new process/utility/people integratedsolutions to laboratory facility design. The requirements forequipment arrangement flexibility, unattended operation,and the critical nature of the data requires newmultidisciplinary interactions in order to provide solutionsthat enhance and enable the overall process in the laboratory.

Adaptability in laboratory design denotes an ability toaccept change. Typically, conventional laboratory designfocused on establishing the parameters of a “laboratoryplanning module.” Traditionally, the laboratory module wasassigned an optimal square footage and provided fixed and/or specialized utilities. Significant cost driven HVAC infra-structure was provided for fixed fume hoods or more flexiblepoint exhausts to support the scientific efforts targeted.Casework was fixed in place and walls became vehicles tosupport and isolate one function from the adjacent laborato-ries. A scientific investment was often measured in thequantity of planning modules assigned to support a specificresearch group.

In today’s robotics oriented discovery laboratory, flexibleopen space is the driver in lieu of the traditional fixed fume

Figure 2. An open plan robotics oriented discovery laboratory.

Page 17: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

Figure 3. A typical automated research laboratory and associated support spaces.

hood, casework, and utility service based laboratory layout.With the robotics oriented discovery laboratory, the mostcommon criteria usually drives an open space environmentwithout restrictive floor service penetrations and one thathas that the ability to support moveable casework/tables thataccommodate highly variable equipment layouts and succes-sive technology platforms. This in turn shifts the investmentfrom initial fixed equipment and casework to mobile researchequipment and the supportive flexible facility infrastructureto serve this equipment.

Figure 1 is a diagram of a traditional chemistry laboratoryfloor plan. It illustrates the fixed equipment and caseworkapproach described previously.

Figure 2 is a diagram of an open plan robotics orienteddiscovery laboratory within the same modular floor plan.This figure illustrates the potential locations of movableequipment and the inherent flexibility of the design.

In new construction of drug discovery facilities, a greaterpercentage of space that would normally be fume hood inten-sive chemistry laboratory is being redefined to accommodateautomated processes. In a recent discovery laboratory project,a full 30 percent of a 140,000 square foot building was devotedto automated research and its associated support spaces.These support spaces typically consist of Plate Storage Li-braries, Chemical Processing Rooms, Cell Reagent BankingSpaces, Cell Culture Areas, Autoclave Rooms, and ColdStorage Rooms. In this recent project, these support spacesrepresented 33 percent of the automated research area.Depending on the customization of the equipment requiredon the site, additional shop areas beyond this example may berequired to construct custom equipment trains. Figure 3

identifies a typical orientation of an automated research labwith associated support spaces.

As laboratories are physically placed higher in buildings,national building codes limit flammable solvent use and itsstorage via the use of control areas. In the past, this hasrequired that chemistry laboratories be placed on lower floorsor designated with a higher hazard classification, whichrequires greater cost and separation from the remainder ofthe facility. Since automated research is typically less solventand waste stream intensive, it can typically be placed higherin the building while maintaining conformance with solventlimitations. In the project illustrated above, all chemicalsynthesis functions were limited to the first and second floorswhile automated discovery laboratories were located on thethird floor.

The researcher population densities of these facilities aregenerally less compared to traditional facilities. Typically,traditional research chemistry buildings are planned at ap-proximately 750 square feet per person. A traditional biologyresearch building generally has a denser population at ap-proximately 600 square feet per person. However, a mecha-nized discovery laboratory is usually planned at1,100-1,200 square feet per person.

Operations with fewer researchers supporting an auto-mated process makes the spatial relationship between thelaboratory and the researcher’s office less critical. The labo-ratory can be located internal to the building and the datacollected, managed, and supported from remote offices. Asthe laboratories can be an unattended operation, spill con-tainment and security are elevated issues. Floor penetra-tions are generally not desired, and in the limited cases where

Page 18: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 4. Mobile laboratory table example.4

they are required, they are typically curbed and sealed. Pastexperience has shown that even when floor penetrations arekept to a minimum they hinder future equipment flexibility.Multiple level card access systems are generally provided tothe laboratory area and supporting sample storage spaces toprovide required security.

The robotics laboratory thrives on open floor plates, wherea limited amount of fixed casework is provided along theperiphery of the laboratory. Optimally, all fixed items such assinks and other equipment are located along perimeter walls.The emphasis has shifted to have the ability to move to othertechnology platforms or alternative equipment arrangementswithout a major infrastructure modification. Mobile, eitherwith or without wheels, refers to something that can bemoved without significant expense, time, or effort. Research-ers will often arrange tables and equipment to best suit therobotic equipment array.

Important to the flexibility of the moveable table conceptis the stability of the equipment once it is in place. Conceptsto achieve space flexibility should thoroughly be tested andproven to achieve the equipment stability required. Forexample, the use of telescoping legs provides flexible benchheight, but may not provide the stability of solid legs. Alter-native mobile casework designs that provide additional sta-bility have been developed. One of these approaches is illus-trated in Figure 44, which is a design that uses a solid legnormally resting on the floor, but by lowering the wheelmechanisms, the table can be fully mobile. Vibration andfloor slab flatness are also prime concerns especially as theassay well count technology continues to rise.

Motorized options also exist to change the height of thework surface in response to individual researcher’s needs.The goal is to take away any barriers that inhibit the scienceand provide an environment that improves productivity. Inthis lab environment, the researchers can make laboratoryequipment, service, and work surface modifications indepen-dently of traditional facilities staff to further increase effi-cient day-to-day operations.

The overall arrangement and size of the HVAC system (as

well as structural floor-to-floor heights and percentage ofmechanical space) in a traditional discovery chemistry build-ing is generally driven by the size, quantity, and simulta-neous usage of fume hoods. The resulting 100 percent outdoorair HVAC system is not only a major first cost driver for theproject, but also an energy consumption driver for thebuilding’s operation. The following HVAC system sizingcomparison can be made for the traditional laboratory designand the automated discovery laboratory design to illustratethe impact of the robotic laboratory trend. The resulting costimpacts are presented at the conclusion of this article.

The typical discovery chemistry laboratory used in thisexample with its associated instrumentation support spaces(Figure 1) required a combination of fume hood types (threebench and one walk in hood was the standard of this particu-lar design). In this laboratory building with 150 of thesehoods, the HVAC system was exhaust driven. From a systemsizing point of view, it is not generally practical to consider allfume hoods open simultaneously as researchers also havesupport spaces and offices in the same facility. HVAC systemsizing was based on an assumed diversified simultaneousoperation of these exhaust hoods to determine a proper andcost-effective HVAC system size. Agreements during designof this discovery chemistry laboratory also limited the work-ing face area of the hood when open, therefore limiting thedesign exhaust airflow required. The floor plan diagram inFigure 1, limiting the working open sash area to 18" in heightand utilizing a room level operating diversity of three out offour hoods open simultaneously, resulted in a maximumroom air change rate of 45 Air Changes Per Hour (ACPH). Aminimum room air change rate of 20 ACPH is required withall hoods closed. Assigning an 80 percent simultaneous op-eration of these labs across the building at design airflows,the total HVAC system is 144,600 cubic feet per minute(CFM) (68.25 m3/sec.). Even with these substantial reduc-tions in exhaust airflow, this equates to 5.15 CFM per squarefoot of lab area (0.026 m3/sec./square meters). In order toreduce this substantial flow to meet actual needs, variable airvolume controls are typically designed for each fume hoodand the overall laboratory to maintain pressurization underthe combination of the variable fume hood sash positions.These control systems have been optimized and refined inrecent years to track actual used airflow based on fume hoodsash positions and limit the system’s outdoor air usage andoperating cost.

The typical automation discovery laboratory illustrated inFigure 2 requires one fume hood in an open 10-module lab.Therefore, the laboratory is not exhaust driven, but coolingload or minimum air change driven. The good engineeringpractice criteria that was applied to these spaces was 10-12ACPH of outdoor air (as solvents and other materials are stillutilized in the lab) and a cooling load of 7-10 watts per squarefoot of cooling load (lights and equipment). These parametersproduce nearly equal design airflows for the space and canreduce the control functions for the space level HVAC systemto a two-position operation. Considering the same area fordiscovery as above, this design produces a system that is

Page 19: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

Figure 5. Overhead service carrier example.5

53,300 CFM (25.16 m3/sec.) and translates to a 1.90 CFM persquare foot (0.0096 m3/sec./square meter) of lab area.

Without the need for fume hood make-up air, adjacentoffices can become an opportunity to return uncontaminatedair to the HVAC system as opposed to providing fume hoodexhaust make-up air. This further reduces the energy cost ofthe laboratory HVAC system.

In addition, more vertical building height is required inthe traditional chemistry laboratory if shafts remain central-ized or alternatively more local vertical shafts are required tobe able to route the required supply and exhaust ductworkand coordinate with other services while maintaining atypical laboratory building structural floor-to-floor height of15-16 feet (4.57 - 4.88 m). With the automated discoverylaboratory, a more centralized shaft arrangement with moretypical building heights can be realized due to reduced air-flow. Therefore, the automated discovery laboratory functioncan offer a substantial building first cost reduction based onthe support systems required.

Given the criticality of the research, the associated opera-tions, the potential for unattended operation, the reliabilityof research equipment, and the associated support systems,supporting utilities are critical. Supporting systems’ designmust begin with this criteria, evaluate all potential modes ofoperation, provide required reliability via redundancy andelimination of single points of failure, and identify monitor-ing requirements to provide reports of upsets to the properpersonnel for prompt response and correction.

With the evolution of laboratory room level control sys-tems and the education of the user in using those systems,there are new opportunities to link the overall HVAC systemoperation to the actual room occupancy. This creates anHVAC system that tracks actual space airflow requirementsbased on actual process usage. The goal is to utilize thebuilding automation system at a user level to minimize theamount of outdoor air that the HVAC system conditions andthus minimize energy costs. When fume hood exhaust vol-ume setbacks (achieved either through variable or constantvolume equipment) were implemented previously, facilitypersonnel typically started awareness programs to makelaboratory users conscious of the costs of leaving fume hoodsopen when not in use. With the focus on reducing energy costsand the variable use of the laboratory spaces, this same typeof user control can be applied to placing the laboratory in theoccupied or full airflow mode only when automation pro-cesses are in use. The HVAC system serving automateddiscovery laboratories is a conventional variable air volumesystem; however, it is the quantity of spaces occupied simul-taneously that creates the system dynamics that tracks theuse of the building. This effort takes the cooperation of theuser, but significant energy savings can be realized. Furtherairflow reductions also can be realized with the enclosing ofthe process equipment and ventilating the enclosed areaalthough consideration always must be given to any openbench work that is required to be completed. The facility canthereby optimize its energy utilization by actually trackingspaces in full operation and only conditioning the amount of

outdoor air required to serve those spaces.As stated previously, the laboratory spaces are to be

flexible in equipment configuration. Therefore, modularityand uniformity in the distribution of piped and electricalutilities is required. One solution that provides flexibility isan overhead service grid as shown in Figure 5.5 Typical pipedservices are compressed air, nitrogen, laboratory vacuum,and carbon dioxide. Service connections are usually quickdisconnects. With the use of adjustable height benches andthe researchers’ ability to make equipment revisions, provid-ing quick disconnects on the overhead service carrier as wellas on the adjustable bench and connecting the services viaflexible hoses allows adjustment of the work surface heightwithout disconnecting the service. It also allows the user toplug-in services as required. Fume hoods also can be mobile.Supply and exhaust trunks also can be thought of as “piped”services.

Power and communications outlets also can be providedon a modular basis in this pre-manufactured service grid tosuite equipment arrangements. The units are accessible torun additional cabling and piping in the future. Floor pen-etrations should be eliminated from the open floor plan asthey limit future equipment layout options. Not only do floorpenetrations limit the open work area, but also these pen-etrations can only accommodate a finite number of utilitylines. Additionally, the disruption to the area below the

Page 20: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

6 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 6. Service columns in areas of larger equipment.

laboratory for future renovation work also adds a furtherdegree of cost and complication. If the floor penetrationscannot be avoided, the opening should be curbed and sealed.The quality of the gas services requires consideration inapplying modular solutions such as cast aluminum manifoldpiped distribution systems. In most applications, these sys-tems provide a flexible solution without impacting the servicequality. However, in some high quality applications, thesesystems should be investigated for their impacts on servicequality. Alternative field constructed service distributiondrops also can be provided. As seen in Figure 66, service dropcolumns can be provided in areas of larger equipment. Alter-natively, a system of cable tray and electrical raceway sys-tems can be utilized for even lower cost solutions.

Cost ImpactsUsing representative current projects in the NortheasternUnited States, the traditional chemistry laboratory facilityas illustrated herein, has a construction cost range of $300-$325/gross square foot of area. Robotic laboratories in thesame circumstances have a construction cost range of $250-$275/gross square foot of area. These ranges reflect buildingand support systems cost only. Laboratory equipment costsare not included as part of these estimates. As stated previ-ously, with a robotics laboratory, cost shifts from infrastruc-ture to laboratory equipment.

Ongoing renovation costs of a traditional laboratory facil-ity include continued relocating of fixed casework. In arobotics discovery laboratory, these renovation costs areprimarily equipment upgrade costs as the spaces are open,the utilities are distributed in a modular organized fashionand generally allow required equipment rearrangementswithout substantial investment in utility reconfigurations.These arrangements of equipment, casework, and utilitiesalso allows for more direct researcher adjustment.

With the yearly cost of conditioning 100 percent outdoorair HVAC systems (in the Northeastern United States) at$4.65/CFM ($9,856/m3/hr.), and the difference in airflowdensity of 3.25 CFM/sq.ft. (0.0164 m3/sec./sq.meter), the sys-

tem serving the automated research space will cost $15.11per square foot less to operate per year than the chemistrylaboratory example described previously. As general airquantities are reduced, the requirements to heat, cool, hu-midify, and reheat make-up air to laboratories (only to beexhausted via a fume hood) also are reduced. A plannedunoccupied mode to reduce air changes when the lab is notoccupied (4 - 6 ACPH) is typically incorporated into therobotics lab HVAC design. This compares with the minimum20 ACPH rate for the example chemistry lab with all hoodsclosed.

Smaller infrastructure systems installed to support thebuilding program result in less maintenance over the life ofthe building. Also control systems serving robotics laborato-ries have fewer dynamic operations, therefore require lesscommissioning and overall maintenance.

ConclusionThe use of robotics in pharmaceutical research is drivingsubstantial and rapid changes in the design of the researchenvironment. These changes range from those associatedwith the provision of flexible furniture and utility systems, tothose involved with the sizing of building engineered sys-tems, and ultimately to the relationship of the laboratory tosupport spaces and researcher offices.

The process of creating laboratory building designs re-sponsive to these changing needs has created a byproduct ofpresenting additional challenges to traditional laboratorydesigns. These challenges are driven by life cycle, not justfirst costs of research laboratories, and require holistic,integrated architectural and engineering approaches utiliz-ing state-of-the-art equipment and designs. They enableresearchers to minimize the limitations of facilities by pro-viding far more direct researcher interaction with supportingsystems and more inherent flexibility in the design. Althougha continuation of trends originating in the past, recent ad-vances in research, equipment, and supporting systems prom-ises to provide rapid progress in meeting these long soughtafter goals.

References1. Caliper 250 Drug Discovery System, on line at

www.claipertech.com/products/hts.shtml.

2. IBM Business Consulting Services Pharma 2005, An In-dustrial Revolution in R&D on-line at www-1.ibm.com/services/strategy/e_strategy/pharmapubindustrial.htm.

3. Price Waterhouse Coopers Pharma 2005 Silicon Rally:The Race to e-R&D on-line at www.fwpharma.com/insightspr99/pharma2005.htm.

4. Figure 4 photo courtesy of Technical Manufacturing Cor-poration (TMC) Peabody, MA.

5. Figure 5 photo courtesy of CASE Systems Inc., MidlandMI.

Page 21: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Robotics Laboratories

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 7©Copyright ISPE 2004

6. Figure 6 photo courtesy of Aventis, Bridgewater ResearchFacility, Bridgewater, NJ.

AcknowledgmentsThe authors would like to thank Mr. Mark Vora of AventisPharmaceuticals, Inc. for sharing his expertise on the back-ground and development of Robotics Laboratories and forproviding access to laboratory facilities and research staff.

About the AuthorsPaul L. Leonard, PE is an EngineeringDesign Principal at Kling. Leonard’s role isto set global engineering criteria and concep-tual engineering direction for the project,considering all goals and objectives, and over-see conformance throughout the project. Hehas 20 years of experience in the develop-ment of systems for research and pharma-

ceutical laboratories, vivarium facilities, pharmaceutical pro-duction facilities, and central utility plants. He has spoken atAHSRAE, ISPE, and Tradelines. He is a member of ASHRAEand ISPE and received his BS in mechanical engineering,Drexel University, 1987. He can be contacted by tel: 1-215/751-2851 or by e-mail: [email protected].

Lou Angelus, AIA is a Principal/ProjectDirector at Kling, where he provides a re-sults-oriented approach to optimize designs,budgets, and schedules. He develops andmonitors project approaches, and he alsosupervises the performance of project teams.As the client’s primary contact, Angelus par-ticipates in project milestones and client meet-

ings to ensure consistency with the client’s goals and needs.He has 20 years of extensive experience in R&D laboratories,vivariums, production, and manufacturing facilities, andcomplex renovation projects. He has spoken at an AAALASNational Meeting. He holds a Bachelor of Architecture fromDrexel University, 1985, and is an active member of both theAmerican Institute of Architects (AIA) and ISPE. He can becontacted by tel: 1-215/569-6144 or by e-mail: [email protected].

Kling, 2301 Chestnut St., Philadelphia, PA 19103.www.kling.us.

Page 22: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

Design Criteria and Evaluation ofPharmaceutical Containment SystemsEvaluation for Closed Isolation Systemsby Osamu Suzuki, PhD, Morihiko Takeda, Koji Tanaka,and Mamoru Numata

This articledescribes thedesign criteriaand acontainmentevaluation forbulkmanufacturingfacilities withclosed isolationsystems.

Background

This article is the sequel to “Design Cri-teria and Evaluation of Pharmaceuti-cal Containment Systems -Evaluationfor Open Isolation Systems” published

in the March/April 2003 issue of Pharmaceuti-cal Engineering. This article describes the de-sign criteria and a containment evaluation forbulk manufacturing facilities with closed isola-tion systems, designed to allow manipulationsat levels of containment below 1 [µg/m3] catego-rized as Exposure Control Limit (ECL) of 4.The results of the containment evaluation un-der the manufacturing processes were comple-mented by a mock-up test to understand thecontainment performances as well as the pre-vious open isolation systems.

IntroductionPharmaceutical containment systems (isola-tion systems) allow pharmaceutical manufac-turers to meet chemical hazardous materialsexposure limits and are a fundamental part ofpharmaceutical current Good ManufacturingPractice (cGMP).1,2 This technology providesunique pharmaceutical containment facilities,including closed and open isolation systems atthe aseptic1-7 or non-aseptic1, 4-7, 9,10 environments.

Pharmaceutical companies, engineering con-tractors, and equipment suppliers have re-cently, but independently, made an effort toestablish the exposure control limit, throughconceptual classification or case studies withvarious surrogate materials.8, 11-15 Design crite-ria for such containment systems in pharma-ceutical facilities was proposed originally in1999.16-19 Since this proposal, the evaluation ofthe containment of potent compounds has beeninvestigated during the design and construc-

tion of several types of pharmaceutical facili-ties with these isolation systems,20-26 including:

• bulk Active Pharmaceutical Ingredient (API)manufacturing facilities

• aseptic dosage form facilities• solid dosage form facilities• multi-purpose facilities

Containment evaluation data using actual po-tent compounds was collected to redress thebalance of:

1. insufficient potent compound containmentevaluation data compared with an abun-dance of surrogate evaluation data

2. insufficient whole facility evaluation data incomparison with an abundance of specificequipment containment data

The purpose of the evaluation of the contain-ment systems is to:

1. investigate quantitatively the containmentlevel of the potent compounds to confirmwhether the containment performances meetthe design criteria proposed

2. establish conclusively the quantitative de-sign criteria based on the analyses of thebehavior of the airborne dust from potentcompounds during manufacturing processes.The previous article described the contain-ment performances in open containmentisolation system that can be categorized as aPerformance-Based Exposure Control Limit(PB-ECL)11 of 3. In contrast, the presentstudy investigated the containment perfor-mances of the closed isolation systems inbulk API pharmaceutical facilities, designed

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 23: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 1. Negative pressurized isolators consisting of filtration isolator with half suits, drying isolator, and dispensing isolator.

to allow manipulations at levels of containment below 1µg/m3 confirming that the performances met the assumeddesign criteria.

Design Criteria for Pharmaceutical FacilitiesThe strategy for the design of pharmaceutical containmentsystems is:16,17

1. Classification of bulk or finished pharmaceutical productsinto five hazard categories according to their inherenttoxicological and pharmacological properties, PB-ECL -Table A. This PB-ECL classification was made on the basisof a previous report11 and previous practices of the phar-

maceutical facilities to contain the various potent com-pounds.

2. Classification of the barrier level into classes from 0 to 2.0,at 0.5 intervals, by integrating various containment sys-tems, including personal protection - Table B. The protec-tion systems for the external environment, such as thelayout for zoning, a HVAC system, and building construc-tion also are considered in the containment facilities.

3. Selection of the barrier level to maintain the requiredenvironment for processing with potent compounds. Thebarrier level should be selected according to the amount ofdust generation depending on the state, such as watercontent and handling volume - Table C.

PB-ECL Category 1 2 3 4 5Exposure Level (µg/m3) 1000-5000 100-1000 1-100 <1 NIL

1. Active Potency (mg/day) >100 10-100 0.1-10 <0.1 <0.12. Hazard Toxicity LD50 (mg/kgRat)

Toxicity of Oral >2000: non-toxic 500-2000: almost 50-500: slightly toxic 5-50: toxic <5: highly toxicnon-toxic

OSHA/HCS >500: non-toxic 50-500: toxic <50: highly toxicWHMIS (Canada) >500: non-toxic 50-500: toxic <50: highly toxicToxic Control Law >300: non-toxic 30-300: slightly toxic <30: highly toxic

Ocean Pollution Control >2000: non- 500-2000: practically 50-500: slightly 5-50: moderately <5: highly hazardous(GESAMP) hazardous non-hazardous hazardous hazardous

Toxicity of intravenous >100: non-toxic 7-100: toxic <7: highly toxic3. Others Carcinogenicity (IARC) - - - 2A, 2B: potentially yes 1: yes

- - -Sensitivity low low-middle middle middle-high high

Table A. Performance-Based Exposure Control Limit (PB-ECL) used for pharmaceutical drug manufacturing.

Page 24: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

4. Integrated study to satisfy the selected barrier level andreflect this in the facility design, together with severalfactors including the conventional barrier technologies,gowning regulation, and layout of building facilities.

Containment Strategy and Descriptionof Processes for Potent CompoundHandling in Bulk Manufacturing FacilitiesPharmacological Properties of the DrugPharmacological properties of the finished product are sum-marized in Table D. Both acute and chronic toxicity areestimated to show severe systemic effect. The pharmaceuti-cal potency is 70 to 150 mg/day. The finished product showsno mutagenic properties, but does demonstrate potentiallycarcinogenic and partially sensitive properties. The finishedproduct can be classified into an ECL category of 1 to 2 (TableA) with respect to the potency and the toxicity, but was placedinto ECL category 4, from the point of view of being poten-tially carcinogenic.

Barrier Level Settingfor each Manufacturing ProcessTable E shows the required barrier level for each process ofthe potent compound handling. The processes include gen-eral procedures in the handling from suspending raw mate-rial, crystallization and filtration, and dispensing for packag-ing. Except for the cake receiving process when wet powder ishandled, level 2.0 was selected as the barrier level for han-dling the potent compound.

Description of Powder HandlingThe potent compounds were totally processed within theclosed system isolation systems. The processes included thefollowing manual operations:

Barrier level Definition for protection regarding worker andenvironment against potent compounds

0 Not protected0.5 Partially protected1.0 Fully protected1.5 More protected2.0 Doubly protected

Table B. Barrier level setting for worker protection.

Table C. Barrier level setting for state of hazard chemicals under ECL.

Barrier levelState Exposure control limit

1 2 3 4 5Large amount of

1.0 1.52.0powder

>Small amount of 1.5 =powder 0.5 2.0Liquid/wet powder 1.0 1.0Very small amount 0.5

0.5 0.5of powder/liquidPowder/liquid 0 0 0 0 0.5enclosedNo hazardous 0 0 0 0 0substances

Figure 2. Plan view of negative pressurized isolators.

Page 25: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Table D. Pharmacological properties of the finished productprocessed.

Properties DescriptionAcute toxicity Severe systemic effectChronic toxicity Severe systemic effectPharmaceutical potency 70-100mg/dayMutagenic NoCarcinogenic Potentially carcinogenicSensitization Partially sensitive

Table E. Process flow and barrier level set.

Process Flow Barrier Level SetSuspending of raw materialCrystallization 2.0FiltrationCake receiving 1.5DryingDispensing 2.0Packaging

1. weighing the raw material within the glove box2. loading the weighed raw material into a reaction vessel

through the inlet which is directly installed within theglove box

3. wet cake filtration on circular pan filter with a 1.2 mdiameter

4. fragmentation of the cake into small pieces and transferinto the neighboring drying vessel

5. dispensing into container and bag-out from the isolationsystem

Details of the isolation system required to maintain thebarrier level of 2.0 are summarized in Table F. The isolationsystems consist of a glove box for loading the material, anegatively pressurized isolator with half suits for the cakefiltration, and isolators with gloves for drying and dispens-ing.

Containment Evaluationof the Isolation SystemsIn the present study, the containment performances wereevaluated in the glove box for loading and the isolators thatconsist of the filtration isolator with half suits, the dryingisolator, and the dispensing isolator. The raw material wastaken into the glove box via the pass box, weighed, and thencharged into the inlet of the reaction vessel, which is directlyinstalled within the glove box to crystallize. The slurryresulting from crystallization was transferred to the isolatorsfor a series of powder handling processes including filtrationfor cake receiving, drying, and dispensing. Figure 1 illus-trates the negatively pressurized isolators. The isolatorswere installed within a prefabricated room where the room

pressure was controlled to be positive with respect to theisolator and negative with respect to the outside environ-ment. This arrangement of the pressure differentials ensuresthe prevention of release of the potent compound from theworking room in unforeseen circumstances. Figure 2 shows aplan view of the isolators. After filtration, the wet cake isfragmented into small pieces manually on the filter by work-ers wearing half-suits in the filtration isolator, and thentransferred to a neighboring drying isolator which is de-signed to be negatively pressurized with respect to the filtra-tion isolator. The dried powder is dispensed into the contain-ers and unloaded using a bag-out system.

Evaluation Methodsa. Evaluation Methods UsedThe evaluation methods used included air sampling of theatmosphere during working and swab testing of varioussurfaces related to the manufacturing processes. A standard-ized method of testing the containment efficiency of solidshandling equipment has been proposed by a working group ofoccupational hygienists and engineers from Europe, the US,and Japan, and is scheduled to be published as an ISPE GoodPractice Guide.27 (The outline of this guide has been pub-lished previously.27) The main purpose of the guidelines is toallow direct comparison of test results both for similar anddifferent types of equipment. However, the guidelines focuson the containment efficiency of equipment using surrogatematerials, rather than the containment performances ofequipment and facilities for actual potent compounds. There-fore, this study referred to the guidelines for the use of themeasurement equipment, but developed the evaluation meth-ods for the containment performances of the installed equip-ment and the facilities.20-26

Figure 4. Containment evaluation in glove box after raw materialcharging.

Figure 3. Schematic view of mock-up booths representing laminarand turbulent flow booths.

Page 26: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

b. Sampling Methods for ContainmentEvaluationAir SamplingThe static monitoring of airborne concentrations in the work-ing environment was performed using air samplers withconstant flow. Airflow volume was calibrated for each airsampler before measurement. A 37 mmΦ-cassette type closedhead was used with a PTFE filter. The cassette head is one ofsampler heads recommended by the Guidelines for Assessingthe Particulate Containment Performance of PharmaceuticalEquipment in addition to an Institute for the OccupationalMedicine (IOM) sampler head.27 Monitoring was started justbefore working and completed just after working. The work-ing times at each process were generally within one hour.Therefore, the airborne concentration was expressed as Short-Term Particulate Airborne Concentration (STPAC), ratherthan long-term PAC by unit of amount collected per unitvolume.

Swab TestsSwab tests were performed for interior and/or exterior sur-faces of some of the equipment and floors of the work areas todetermine the degree of surface contamination. A previouslyreported28 standardized swab procedure was used with minormodifications. Commercially available cellulose cloth, 10 cmsquare, was dipped in purified water, squeezed, and thenused for swabbing. The swabbing was performed by the sameperson throughout. Surfaces such as the interior and/orexterior of isolator, floors around isolators, and the degowningroom were included in the swab tests. The concentration wasexpressed either as total amount recovered or as amount perunit surface area.

Measuring PointsMeasuring points for air sampling and swab tests wereselected to incorporate:

1. interfaces between containment equipment and workingenvironment

2. final points where the airborne dust collected, namely, infront of exhaust

3. floor surfaces traversed by the workers to account for carryover of potent compounds

These points take into consideration both the evaluation ofequipment efficiency and the overall facility containmentperformance.

Analytical MethodThe concentrations of airborne and surface-contaminatedparticulate were analyzed by chemical analysis using HPLC.The recovery rates from both the filter and the cloth werevalidated and reflected in the analytical results. While therecovery rate from the surface was not determined, thesurface was visually checked to confirm that no powderremained after swabbing. A previous study using this swabmethod28 confirmed the recovery rate of 95% confidence of thetest material from the surface although an organic solventwas used for dipping swab cloths in this earlier study.

Containment Facilities with Processing

Glove boxRaw material loading into the reaction vessel(Highly potent raw material to be loaded into the reaction vessel)

Isolator with half suitesFiltration(Wet cake to be collected by half-suite operators)

Isolator with glovesDrying/dispensing(Dried powder is dispensed into the containers)

Table F. Containment facilities used in the facilities.

Containment Facilities Description

Box pressure to be controlled negative1. Raw material containers to be loaded through the pass box2. Raw material to be loaded into the reaction vessel in this isolator3. HEPA filter to be amounted for air supply/exhaust line4. Waste matter to be unloaded from the isolator by means of bag-out system

Chamber pressure to be controlled negative1. New parts/filter/containers to be loaded through the pass box2. Intermediate products to be transported into the drying/dispensing isolator

directly3. HEPA filter to be amounted for air supply and exhaust4. Waste matter to be loaded from the isolator by means of bag-out system.

Chamber pressure to be controlled negative1. Containers to be unloaded with bag-out system2. HEPA filter to be amounted for air supply and exhaust3. Waste matter to be unloaded from the isolator by means of bag-out system

Figure 5. Interior surface contaminant concentrations of glove boxafter raw material charging.

Page 27: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

6 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

c. Mock-Up TestParameters for the Mock-Up Test andDescription of the Mock-Up BoothMock-up test covers several parameters that should be con-sidered in the design of isolation systems. They includestudies of the isolation systems, such as whether it is closedor open; flow patterns, whether they are laminar or turbu-lent; efficacy of local exhaust ventilation; different types ofthe surrogate materials and the particle size distributions;and type of sampler head, such as the cassette type closedhead or the IOM head for air sampling.

Figure 3 shows a schematic view of the mock-up boothrepresenting laminar down flow and turbulent flow boothswith local exhaust. A dust feeder was used to secure constantfeeding of the surrogate material. In this study, the compari-son between open and closed booths with laminar flow as aparameter was performed using a surrogate material. Thecassette type closed head was used for air sampling.

Surrogate MaterialAcetaminophen was used as surrogate material with a par-ticle size range of below 75µm by weight ratio of 66% (pro-vided by the manufacturers and determined by sieving num-ber).

Results and DiscussionEvaluation of Containment Performancesof the Bulk Manufacturing FacilitiesGlove Box Interior and Exterior Surfaces,and Floors around Glove BoxFigure 4 shows the sampling points and the analytical resultsfor both air sampling and swab test. This figure is illustratedin plan view. Air sampling was performed out in front of thepass box where the potent compounds were collected. Swabpoints included floors in front of the glove box and around thereaction vessel that covers areas of work with swab areas of1.3 m2 and 0.7 m2, respectively. The results in both analysesfor air sampling and swab test indicated that concentrationsof the potent compounds recovered were below the detectablelimit of the chemical analysis used. Figure 5 shows theresults for the interior surface concentrations of the glove boxafter the raw material weighing and charging processes. Theconcentration of the interior surface contaminations arepresented, expressed by a unit of amount recovered per 100cm2 that corresponds to the swab area. The funnel was usedwhen charging raw material into the reaction vessel. Theshaded areas correspond to the areas where local exhaustswere installed. The results indicated that surface concentra-tions generally increased downward, which may reflect air-borne concentration distribution inside of the glove box. Theflow within the glove box of the airborne dust could typicallybe controlled by the airflow. The major part of airborne dustcould be collected by the HEPA filter of the exhaust. However,the large particles within the airborne dust seem to dropdown within the glove box. In contrast, the exterior surfaceconcentration from all front surfaces that were swabbed wasbelow the detectable limit (<1µg/m3). These results demon-strated that the potent compound was fully contained at alevel below the detectable limit of the chemical analysis used.

Isolators for Filtration, Drying, and DispensingFigure 6 shows the sampling points and the analytical resultsfor the isolators. Air monitoring was performed:

• in front of the filtration isolator between the unloading boxfor the filter and the pass box for tools and parts

• in front of the bag-out port of the dispensing isolator• in front of exhaust in the room

The areas swabbed included floors in front of the filtrationisolator and the dispensing isolator and the floor of thedegowning room. These sites cover areas traversed by work-ers.

The dispensing process was performed the day after thefiltration and the drying processes. The swabbing in front ofisolators was performed after each process was completed.The swabbing of the degowning room was performed after thefiltration and drying processes. These sites covered areastraversed by workers both during and after working.

The results revealed that the analytical concentrationswere below the detectable limit (<1[µg/m3]) for both airsampling and swab tests during the filtration and the drying

Figure 7. Containment evaluation in negative pressurized isolatorsafter dispensing process.

Figure 6. Containment evaluation in negative pressurized isolatorsafter filtration and drying processes.

Page 28: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 7©Copyright ISPE 2004

Figure 8 shows the STPAC of airborne acetaminophen for abase case (no airflow case), a local exhaust working case, andfor closed system with laminar flow in the mock-booth. Thenumerical values are shown for both inside and outside thebooths. The results demonstrated the quantitative efficacy ofthe local exhaust to reduce the STPAC (from 7,000 [µg/m3] to1,000 [µg/m3] for inside and 1,000 [µg/m3] to 10 [µg/m3] foroutside the booths) and the containment performance of theisolation system to contain the surrogate material fully (66[µg/m3] for inside and below detectable limits for outside thebooth). These results also could provide a theoretical back-ground to understand the containment performances of thebulk APIs manufacturing facilities evaluated in the presentstudy. Recent Computational Fluid Dynamics (CFDs) stud-ies suggested that the exhaust has an important role tocontain the compound by design of the airflow in the transfervessel having a local exhaust system.2 In such an openisolation system, it appears that understanding the airflow iscritical to ensure the containment performance. On the otherhand, it has been shown that particle size distributions ofairborne dusts were different depending on the state of thepowders despite using the same materials,29 suggesting thedifferent behavior of airborne distribution. Further study atthe various conditions with different surrogate materials isrequired to fully understand the containment performancesof the closed and open isolation systems.

ConclusionFrom the measurement of containment performances of theglove box and the isolators, the present study can be summa-rized as follows:

processes. Figure 7 shows that the containment performancesachieved analytical concentrations below the detectable limit(<1[µg/m3]) for both air sampling and swab tests during thedispensing process.

These results demonstrated that the potent compoundsprocessed in the isolators were fully contained at a levelbelow the detectable limit of the chemical analysis used. Inaddition to the results obtained with the glove box, thecontainment performance was shown to meet assumed de-sign criteria for both the isolation equipment and the wholefacility, including the isolators.

A Study of Containment EvaluationIt is important to confirm that the containment performancesof equipment and facilities meet the design criteria. Thisstudy demonstrated that the containment performances ex-amined achieved analytical concentrations below the detect-able limit of the chemical analysis used. However, it shouldbe noted that the containment performance is influenced bythe operational procedure. Previous studies suggested that acontaminated gown may be one source for carrying anypowder handled outside the working area although in thiscase, the facilities evaluated included open isolation systemsand were classified as ECL 3, which allows an exposure levelbetween 1 and 100 [µg/m3].20-26 The manufacturing processesalong with standard operating procedures would be recom-mended from the point of view of total security of containmentfor the potent compound handling, even where the closedisolation systems are used.

Containment Performance Evaluation byMock-Up Test

Figure 8. Containment evaluation in mock-up booth for open and closed isolation systems.

Page 29: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

8 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

1. The containment performances of bulk APIs manufactur-ing facilities, designed to allow manipulations at level ofcontainment below 1 [µg/m3], were confirmed to meetassumed design criteria.

2. The closed system was compared to the open system usingthe mock-up booth to complement the performances of thebulk manufacturing facilities, demonstrating quantita-tive containment performances with non-detectable lev-els.

3. The influences of airflow and local exhaust on contain-ment performances were delineated, and although frag-mentary, they were quantitative.

Further studies are under way to understand systematiccontainment performances in different kinds of control sys-tem, such as type of airflow, using the mock-up system.

References1. Belly, S. and J. Wilkins, “A Technical Review of Isola-

tors,” Pharmaceutical Engineering, Vol. 18, No. 2, 1998,pp. 80-87.

2. Ramsden, A., “The Use of Airflow Modeling in the Designof Pharmaceutical Containment Systems,” Pharmaceuti-cal Engineering, Vol. 21, 2001, No. 4, pp. 40-46.

3. Friedman, R., “Design of Barrier Isolators for AsepticProcessing: A GMP Perspective,” Pharmaceutical Engi-neering, Vol. 18, No. 2, 1998, pp. 28-36.

4. Francis, L., “The History and Principles of Barrier Tech-nology for the Pharmaceutical and Biotech Industries,”Pharmaceutical Engineering, Vol. 18, No. 2, 1998, pp. 38-47.

5. Madsen, E. R., “Design and Validation of Isolator Sys-tems for the Manufacturing and Testing of HealthcareProducts. Highlights from the New PDA Technical ReportNo. 34,” Pharmaceutical Technology, Vol. 25, No.2, 2001,pp. 40-44.

6. Lysfjord, J. and M. Porter, “Barrier Isolation History andTrends, A Millennium Update,” Pharmaceutical Engi-neering, Vol. 21, No. 2, 2001, pp. 142-148.

7. Lysfjord, J. and M. Porter, “Barrier Isolation History andTrends,” Pharmaceutical Engineering, Vol. 18, No. 5,1998, pp. 80-86.

8. Liberman, D., C. Lockwood, M. McConnell-Meachen, E.McNally, H. Rahe, K. Shepard, and G. Snow, “BarrierIsolation Technology: A Safe and Effective Solution forProviding Pharmaceutical Development Facilities,” Phar-maceutical Engineering, Vol. 21, No. 4, 2001, pp. 22-30.

9. Hines, M. J., “Conceptual Design of Isolators for Han-dling and Processing Dry Bulk Pharmaceutical Chemi-cals,” Pharmaceutical Engineering, Vol. 18, No. 2, 1998,pp. 60-78.

10. Guest, I., “Industrial Hygiene Challenges Presented bythe Trends in New Chemical Entities,” ISPE Confer-ence, Amsterdam, Netherlands, December 8-9, 1999.

11. Naumann, B.D., E.V. Sargent, B.S. Starkman, W.J. Fraser,G.T. Becher, and G.D. Kirk, “Performance-Based Expo-sure Control Limits for Pharmaceutical Active Ingredi-

ents,” American Industrial Hygiene Association Journal,Vol. 57, 1996, pp. 33-42.

12. Harrison, D., “Matching Containment Equipment to theproduct Hazard,” ISPE Conference, Amsterdam, Neth-erlands, December 8-9, 1999.

13. Ryder, M., “Design Requirements Criteria for Contain-ment Device Selection,” ISPE Conference, Amsterdam,Netherlands, December 8-9, 1999.

14. Harrison, D., “Validating the Performance of Engineer-ing Control Systems,” ISPE Conference, Amsterdam,Netherlands, December 5-6, 2001.

15. Koch, M., “Standard Testing of Containment of Connec-tion Systems,” ISPE Conference, Amsterdam, Neth-erlands, December 5-6, 2001.

16. Takeda, M., “Trends in Handling Potent Compounds inJapan,” ISPE Conference, Amsterdam, Netherlands,December 8-9, 1999.

17. Tahara, S. and K. Watanabe, “Construction of DesignProcedure for Chemical Hazard Facilities. Part 1 of 3,”Pharm Tech Japan, Vol. 15, No.10, 1999, pp. 1441-1450,in Japanese.

18. Takeda, M., K. Ito, O. Shirokizawa, and K. Watanabe,“Construction of Design Procedure for Chemical HazardFacilities. Part 2 of 3,” Pharm Tech Japan, Vol. 15, No.11,1999, pp. 1619-1630, in Japanese.

19. Masuda, M., M. Kawai, K. Tozaki, and K. Watanabe,“Construction of Design Procedure for Chemical HazardFacilities. Part 3 of 3,” Pharm Tech Japan, Vol. 15, No.12,1999, pp. 1769-1780, in Japanese.

20. Suzuki, O., “Design Criteria for Barrier Isolation Systemsand Containment Evaluation in Pharmaceutical Facili-ties,” ISPE Conference, Amsterdam, Netherlands,December 5-6, 2001.

21. Suzuki, O., “Design Criteria and Evaluation for Pharma-ceutical Containment Systems,” ISPE Conference,Philadelphia, USA, September 11-12, 2002.

22. Suzuki, O., “Design Criteria and Evaluation for Pharma-ceutical Containment Systems,” ISPE Conference, Ber-lin, Germany, December 2-3, 2002.

23. Suzuki, O. and M. Takeda, “Evaluation of ContainmentPerformances in Pharmaceutical Containment Systems.Part 1. Evaluation for Open Isolation Systems in BulkManufacturing Facilities,” Pharm Tech Japan, Vol.18,No.11, 2002, pp. 31-39, printed in Japanese.

24. Suzuki, O., K. Tanaka, W. Naruaki, and M. Takeda,“Evaluation of Containment Performances in Pharma-ceutical Containment Systems. Part 2. Evaluation forOpen Isolation Systems in Dosage Form ManufacturingFacilities,” Pharm Tech Japan, Vol.19, No.3, 2003, pp. 59-68, printed in Japanese.

25. Suzuki, O., M. Takeda, K. Tanaka, and M. Inoue, “DesignCriteria and Evaluation for Pharmaceutical Contain-ment Systems. Evaluation for Open Isolation Systems,”Pharmaceutical Engineering, Vol. 23, No.2, 2003, pp. 66-78.

26. Suzuki, O., K. Tanaka, W. Naruaki, and M. Takeda,“Design Criteria and Containment Evaluation for Phar-

Page 30: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Containment Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 9©Copyright ISPE 2004

maceutical Containment Systems in Aseptic Dosage FormManufacturing Facilities,” Pharmaceutical Technology,Aseptic Processing 2003, pp. 24-30.

27. Gurney-Read P, Koch M., Guidelines for Assessing theParticulate Containment Performance of Pharmaceuti-cal Equipment. Pharmaceutical Engineering, Vol. 22, No.3, 2002, pp. 55-59.

28. Cocker, N., “Vale III Isolation Validation - GuaranteedProtection,” ISPE Conference, Amsterdam, Nether-lands, December 8-9, 1999.

29. Pujara, C.P. and D.O. Kildsig, “Effect of Individual Par-ticle Characteristics on Airborne Emissions,” Contain-ment in the Pharmaceutical Industry, Drugs andPharmaceutical Sciences, Vol. 108, ed., Wood, J.P.,Marcel Dekker, Inc., 2001, pp. 29-53.

About the AuthorsOsamu Suzuki is Professor, Division of Cran-iofacial Function Engineering, Tohoku Uni-versity Graduate School of Dentistry. Hejoined the university in March 2004 from theResearch and Development Center of JGCCorp. He has studied surface and interfacesciences in the field of biomedical science,petrochemical and pharmaceutical engineer-

ing. He received his PhD from Tohoku University GraduateSchool of Medicine in 1991; was Visiting Scientist at ForsythResearch Institute, Physical Chemistry Department (Bos-ton, MA, USA) from 1992 to 1994; Research Associate inJapan Fine Ceramics Co. Ltd, a subsidiary of JGC from 1986to 1994; moved to JGC Corp in 1995 and was a SeniorResearcher from 1999 to 2004. He has been an ISPE membersince 2001. He can be contacted by e-mail: [email protected].

Tohoku University Graduate School of Dentistry, Cranio-facial Function Engineering (CFE), 4-1, Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan.

Morihiko Takeda is President of Pharma-ceutical Solutions Corp. He assumed the lead-ership role in the containment evaluationresearch project at JGC Corp. He proposedthe design criteria for the pharmaceuticalcontainment facilities described in this ar-ticle. He was a speaker at the 1999 ISPEAmsterdam Conference, Containment of Po-

tent Compounds. He has been acting as lead engineer tovarious projects for parenteral and solid dosage productsmanufacturing plant for domestic and multinational compa-nies. He has BSc in mechanical engineering from SaitamaUniversity in 1981. He joined JGC in 1981. He foundedPharmaceutical Solutions Corp., Engineering Company dedi-cated for containment and isolation technologies in 2004. Hehas been an ISPE member since 1999. He can be contacted bye-mail: [email protected] or by tel: +81-45-543-3938.

Pharmaceutical Solutions Corp., Ltd., 664-2 Mamedo-cho,Kouhoku-ku, Yokohama 22-0032, Japan.

Koji Tanaka has been Research Associatein the Research and Development Center ofJGC Corp. since 1999, and has been dedicat-ing himself to obtaining and analyzing thepharmaceutical containment evaluationdata. He has experience in analyzing variouscontainment systems including nuclear fa-cilities that are required to keep the reduced

condition inside the glove box. He learned various handlingmethods for hazardous chemicals, poisonous gases, and ra-dioactive substances through these containment evaluations.He has an MSc in analytical chemistry from Kyoto UniversityGraduate School of Engineering, Department of MolecularEngineering by studying syntheses and properties of organicconductors in 1999. He has been an ISPE member since 2001.He can be contacted by e-mail: [email protected].

JGC Corp., Research & Development Center, 2205 Narita-cho, Oarai-machi, 311-1313 Japan.

Mamoru Numata is Manager of Technol-ogy Development Department in the Researchand Development Center of JGC Corp. Hereceived MSc from Tokyo Metropolitan Uni-versity in 1978 and joined JGC Corp. in 1978;was visiting researcher at Belgian NuclearResearch Center from 1984 to 1985; visitingresearcher at Japan Atomic Energy Research

Institute from 1987 to 1989; Senior Principal Researcherfrom 2001 to 2003 and Manager of Technology DevelopmentDepartment in Research and Development Center of JGCCorp. from 2004. He has been a member of the JapaneseSociety for Bone and Mineral Research, the Japanese Asso-ciation for Oral Biology, the Chemical Society of Japan, theAtomic Energy Society of Japan, and the Japan Society ofNuclear and Radio-chemical Sciences. He can be contacted bye-mail: [email protected].

JGC Corp., Technology Development Department in Re-search & Development Center, 2205 Narita-cho, Oarai-machi,311-1313 Japan.

AcknowledgementsThe authors are deeply indebted to Engineer Jun Sugiura,Group Leaders Tetsuo Fujioka and Shinichi Yamagami,Senior Managers Makoto Miyamoto, Keiichiro Watanabe,and Shinichi Osada who all are members of the IndustrialProject Division, JGC for their helpful suggestions and valu-able criticism during the evaluation of the containmentfacilities and the preparation of the manuscripts. The au-thors also thank Michiko Ishizuka of Industrial Project Divi-sion, JGC for her excellent support in improving the manu-script.

Page 31: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Clinical IVR Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

An Introduction to Clinical IVRSystemsby Anthony Chow

This articlepresents waysin which ClinicalInteractiveVoice Response(IVR) Systemsare used toenhance trialcommunications,request trialsupplies forinvestigationalsites, provideinstant updatesfor trialsponsors, andsave money inthe process.

Overview

Extremely reliable, cost-effective, andeasy to use, Clinical Interactive VoiceResponse (IVR) Systems provide a widerange of services from just about any

telephone worldwide. Successful integration ofan IVR system with a clinical trial involves anunderstanding of the capabilities of the tele-phone as a data collection device, and beingable to utilize secondary systems, such as fax,e-mail, and web portals, to post and routeimportant information. This article will ex-plore some of the many ways in which IVR isbeing used to enhance trial communications,request trial supplies for investigational sites,provide instant updates for trial sponsors, andsave money in the process.

A Brief History of Clinical IVRDuring the last decade of the 20th century, theuse of IVR systems for clinical trials grew, notjust in number, but also in variety. Most clini-cal teams are familiar with using IVR systemsto capture patient diary data or to randomizetrial subjects. Due to the many capabilities ofIVR systems, and the fact that almost everyonehas a telephone these days, IVR systems havebeen developed to automate more and more ofthe clinical trials process. IVR systems havebeen developed for investigators to order studydrugs and trial supplies. Sponsors can placetoll free numbers in radio and newspaper adsthat permit callers to enroll in upcoming trialsvia IVR. Some systems even serve as informa-tion centers telling callers about particulartrials and drugs.

As IVR systems took on more responsibili-ties, the capabilities of Clinical IVR Systemshad to be expanded to meet the demand. Statis-tics and reports detailing IVR activity may beautomatically faxed or e-mailed to clinical team

members. Compliance reports assist investiga-tional sites in reminding trial subjects to makediary entries. IVR randomization systems cansend faxes to drug shipment facilities. Poten-tial trial subjects can enter a fax number intoan IVR Subject Recruitment System and re-ceive study details. Data collected by IVR sys-tems can be made available to most anyoneneeding such information.

The reach of IVR for clinical trials has be-come global with the addition of foreign lan-guages. Many language translation vendorsnow provide recording services for IVR sys-tems. Some are even familiar with the pharma-ceutical/biotech industry and can assist withvalidation of IVR systems.

The potential power and reach of any Clini-cal IVR System is as vast as the global telecom-munications network in existence today. Inaddition to voice recognition, newer IVR sys-tems hardware can even decode digits from theold rotary-dial telephones. All of this evidenceproves that Clinical IVR Systems are the mostreliable form of electronic data capture in exist-ence today for the pharmaceutical/biotech in-dustry.

How does it Work?When gathering information on IVR systems,many people are curious as to how the systemsfunction. Most IVR systems are hosted on ordi-nary PC workstations outfitted with specialtelephone circuit boards, or add-on cards. (Thereis support for hosting IVR on multiple plat-forms such as Windows, Unix/Linux, andSolaris.) The telephone cards can accept two ormore telephone lines, and in function, are dif-ferent than fax/modem cards. Some telephonecards can accept an entire T-1 ISDN PRI net-work interface containing 24 lines. With mod-ern PCs getting more powerful each year, it is

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 32: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Clinical IVR Systems

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

not unusual to find more than one phone card in any givenworkstation. Being able to handle many simultaneous call-ers, e.g., 64 lines, in a single box makes IVR a very economicalinvestment.

Touch-TonesThe main job of a telephone card is to translate the touch“tones” into numeric digits. When a key is pressed on a touch-tone telephone, a sound is generated comprised of two tones.One tone represents the key’s row; the other represents thekey’s column. Together they represent digits 0 through 9, *,and #.

Voice FilesThe second job of the telephone card is to “play” prerecordedprompt files to the caller. The caller “interacts” with thesevoice prompts, which is how this system got its acronym,Interactive Voice Response, or IVR. For example, when call-ing an IVR system, you may first hear a greeting followed bya request for your user ID and password. Your “response” isto enter the ID and password. Each of these prompts may bea separately recorded voice file on the IVR workstation. Voicefiles may be stored as WAV1 files, or more commonly, as VOX2

files.

IVR ScriptsThe two basic functions described above, playing prompt filesand receiving touch-tone responses, are controlled by one ormore IVR scripts or call flow. The scripts are processed by themain IVR program and contain the business logic of theapplication. An IVR scripting language may take many formsincluding XML. Basically, the IVR script contains the neces-sary information to prompt callers for information and storetheir responses.

IVR scripts are often role-based by design. For example,when a site investigator logs in, the IVR system may presentthe caller with administrative options, such as patient with-drawal and activating/de-activating an investigational site.Whereas, when a patient calls, the IVR system may onlyallow access to the patient diary options.

To improve data integrity and reliability, IVR scriptsusually include repeating the data entered so that callers canhear and confirm their selections. For example, a patientfrom the UK calls in, whose birthday is August 1, 1946:

IVR plays the prompt: Please enter your birthday using 2-digit month, 2-digit day, and 4-digit year. For example,January 9th, 1950, will be 01091950.

Caller punches in: 01081946

IVR repeats: You have entered January 8th, 1946. If it iscorrect, press 1. Otherwise, press 2 to enter again.

Caller hears the mistake and punches in: 2

IVR plays the prompt: Please enter your birthday using 2-digit month, 2-digit day, and 4-year year. For example,January 9th, 1950, will be 01091950.

Caller punches in: 08011946

IVR repeats: You have entered August 1st, 1946. If it iscorrect, press 1. Otherwise, press 2 to enter again.

Caller confirms the entry and punches in: 1

Basic Clinical IVR Custom Programming Requirements

If your company or facility is interested in hosting a Clinical IVR System, andyou have access to a programmer, there is a very economical way to getstarted. Here are some basic hardware and software requirements.

1. PC Workstation (Windows 2000 or later preferred)2. 2 or 4-line Dialogic Card3. CallSuite software (2 or 4-line license)4. Microsoft Visual Basic (Version 6 or .NET)5. Microsoft Access, SQL Server, or Oracle database

Using very simple logic, and basic database access routines, a very useableClinical IVR System can be designed and developed. In fact, the mostdifficult part of the process for most first-time programmers is getting thedrivers installed for the Dialogic1 hardware. Once the system is “talking,”and can answer incoming calls, the hard part is mostly over.

Getting StartedThe first routine to develop will be a utility to permit recording of voiceprompt files. With a little ingenuity, a simple scripting system can bedesigned to read voice prompt filenames from a table in your database. Thebasic functionality needed for a recorder routine can be seen in the followingmenu prompt, which will be the first file you need to record.

“To record a file, press 1. To playback, press 2. To move to the next file,press 3. To move to the previous file, press 4. When finished, you may hangup.”

Therefore, you will need to record a file, play a file, and move up or downthrough a list of voice files. Just coding this simple routine will give youvaluable experience in how all IVR systems function. For example, usingsimilar logic, a function can be written to permit Investigators to leave avoice message for someone on the sponsor’s clinical team.

To improve on the functionality of this routine, error checking can beadded to prevent playback of files that do not yet exist. You can also addchecking to assure the caller does not move the index beyond the bounds ofthe record object, or array, of filenames being used. The experience gainedwith this simple exercise can lead to development of a fairly sophisticatedscripting engine to handle Clinical IVR scripts.

Error HandlingProbably the most important aspect of a Clinical IVR Visual Basic program isto incorporate extensive error handling in each function and/or subroutine.The system must detect when a caller hangs up. By default, in Visual Basic,this generates an error. This error must be trapped and handled so you canfinish writing any data to your database. If the error is not trapped, programcontrol may otherwise bypass the database write routines resulting inincomplete or inaccurate information.

ReferenceThere are many computer telephony vendors that can provide IVR worksta-tions with telephony hardware preinstalled and configured, ready for usewith most telephony software systems.

Page 33: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Clinical IVR Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

DatabaseWorking in conjunction with the IVR script will be some typeof database. The IVR system database is used to store andretrieve scripts, user ID and password information, and datacollected from the callers. Additional information on IVRusers may be stored in the database, such as contact informa-tion and reports preference.

More important, the database stores electronic records foraudit trails per FDA 21 CFR Part 11, capturing each user’sevents and actions along with user ID and timestamps.

The same IVR system database, or perhaps a separatedatabase, can serve as version control to track the design ofvoice files and IVR scripts.

IVR PackagesThere are many software packages on the market that permityou to generate IVR scripts, record voice files for prompts,and capture data. Some IVR packages allow you to build anIVR script visually and interactively using icons and flowconnectors.

Custom ProgrammingTo custom build a Clinical IVR System, there are severalthird party programming libraries, or toolkits, available formany computer languages. One of the more common librariesencountered is CallSuite, which is a set of commercial ActiveXcontrols with computer telephony features. CallSuite may beused with Visual Basic, Visual C++, or Delphi.

These programming libraries provide transparent accessto the telephony hardware. They give you control over whatfiles are played and how they are played, and the ability tocapture digits entered by callers into program variables. Youalso can write code to record a voice message from the caller,change the length of time the system will wait for a caller’sresponse, and also put the caller on hold and forward them toanother extension or another telephone number. Basically,you can duplicate just about any type of telephone systemusing these routines. As an example, many cable TV compa-nies use computer telephony hardware with Visual Basicprograms to process calls from customers wanting to pur-chase “pay per view” movies. These systems are able to readthe Caller ID information sent between the first and secondrings and activate the selected movie for the customer auto-matically.

In addition to the advantages above, it is possible to codeextra functionality into your IVR system. For instance, youcan easily build a transaction log of all caller activity and thensave or “publish” the information. You can code routines tosend faxes or e-mails to staff members. If a caller is havingtrouble, you can give them an option to talk to a live operator.The real power of Clinical IVR Systems is in these auxiliarysystems that provide people easy access to information.

A Survey of Clinical IVR SystemsSubject DiariesOne of the more common types of Clinical IVR Systems,subject diary, is used by many drug companies to corroborate

the study results collected from investigational sites. Drugefficacy, alone, is no indication that a drug will succeed in themarket place. The patient’s “quality of life” while on studymedication must be at acceptable levels as well.

Clinical IVR diaries have been hugely successful for manyreasons, but mostly because of the excellent quality of thedata collected. This is mainly due to the fact that most IVRdiaries do not permit subjects to make entries for past days.In other words, subjects only have 24 hours in which to entertheir “daily diary.” This prevents subjects from cheating.Many investigators have observed trial subjects filling outtheir “daily diaries” on paper for the past two weeks whilesitting in their waiting room. And, just recently, one investi-gator caught a subject pre-filling in their diary data for thenext two weeks! Unlike paper diaries, which are held in thesame regard as homework, Clinical IVR Systems providesimple constraints that drastically improve data quality andvalue.

Another major factor that can impact data quality istaking some extra time to design the diary script expressly foran IVR3 System. IVR systems are live 24 hours a day. There-fore, subjects can be instructed to make diary entries at theonset of a significant event. For example, when a headache isstarting, or pain relief is indicated, subjects can call thesystem and enter data on how they are feeling “right now.”This fact alone makes IVR diaries superior to all otherformats, especially paper. Having immediate access fromvirtually any telephone, plus the constraints on enteringinformation only for “today,” combine to produce accurateand timely diary data.

Randomization SystemsClinical IVR Systems have been used to randomize subjectsfor blind, double blind, open-label, and virtually any othertype of clinical trial. A true central randomization can beachieved by using a central computer to select the appropri-ate treatment, which is a feature most statisticians find veryattractive. The randomization number, as well as the treat-ment name, can be played to the caller over the telephone sothey can act on the information immediately. Once therandomization is completed, a confirmation report can begenerated and either faxed or e-mailed to the investigationalsite to file with their records.

Trial/Drug Supplies, Inventory, andRandomization Systems

Using another very popular type of Clinical IVR System,investigational sites can call and request study drug and/ortrial supplies while permitting the sponsor’s clinical team tomonitor inventory and supplies. When combined with arandomization feature, such IVR systems function as theheart of a clinical trial. Reports and statistics on active sites,enrolled/randomized subjects, and supply usage can easily begenerated and automatically sent to the appropriate clinicalteam members. The easy addition of a Completion/Withdrawscript completes the system and provides pertinent details onthe trial.

Page 34: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Clinical IVR Systems

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Common Auxiliary ServicesMerely collecting such information does little to enhance atrial. The real power and value of Clinical IVR Systems is inhow such information is presented and distributed.

Fax/E-mailA successful Clinical IVR System must include reports andstatistics and an easy method for users to receive and viewthem. Most people have e-mail addresses these days. But, notall biopharmaceutical companies are permitting informationto be sent in this way. For Clinical IVR Systems, the mainvenue for reports, confirmations, and summary statistics isstill the fax.

The following is a partial list of the many reports that canbe produced from Clinical IVR Systems:

1. Active Sites2. Global Subject Enrollment3. Subject Enrollment by Site4. Subject Enrollment by Treatment (unblinded only)5. Completed Subjects6. Discontinued Subjects7. Summary of Completed/Discontinued Subjects8. Trial Supplies Inventory9. Trial Supplies Alerts (when to order new supplies)10. Trial Supplies Usage by Site11. Study Drug Shipments12. Unused Study Drugs

Keep in mind that, for information to be reported, it must becollected. Fortunately, Clinical IVR Systems make collectingsuch information a very easy task.

As with the information itself, the reports are of little useunless they can be distributed to the right people in a timelyfashion. Incorporating faxing and e-mail capabilities intoreport generation systems is not uncommon for most pro-grammers these days. Once a report has been generated as aresult of a pre-defined event such as a randomization event,scheduled report generation, or a direct request from anInvestigator, it is a simple matter to call the appropriatefunction to fax or e-mail the report to the appropriate recipi-ent. Some Clinical IVR Systems can even publish reports andsummary statistics on report usage, which provides usefulmetrics for future trials.

Web PortalWeb portals, or interfaces, usually reserved for clinical teammembers, can permit fast access to all Clinical IVR data andprovide forms for initializing new investigational sites, orupdating fax/e-mail information. In addition to providingaccess to information, a web portal can allow users to managemore of the project.

For example, when activating accounts for a site for a largetrial, the site administrator may need to provide the IVRsystem administrator a long list of site user’s names, faxnumbers, and e-mail addresses via paper, fax, and/or e-mail

so the IVR system administrator can enter the information.Instead of this manual paper-handling process, a web portalcan provide clinical team members with the ability to enterand update information to save time and effort.

Web portals also can provide the clinical team with admin-istrative capabilities for managing key events in a clinicaltrial such as granting waivers, or exceptions. For example,some patients may not be eligible for randomization into thetrial due to multiple entrance criteria based on the number ofdiary entries made, as well as the number of clinicallysignificant events occurring in a given period. A clinical teammay use the IVR system administrative capabilities to grantwaivers, thus allowing these subjects to be randomized intothe trial.

Hand-Held Computer DevicesThe emergence of hand-held computer devices as a means ofdata capture is noticeable in clinical trials today. Working inconjunction with IVR systems, hand-held computer devicesare useful to capture sensitive, text-based patient data.However, distributing these devices to all study subjects canbe costly. Also, they can easily be misplaced and lost. There-fore, hand-held computer devices by no means should replaceIVR systems, but may improve data reliability if they areboth available.

Saving Money on Trial Supplies andStudy Medication

One of the more recent developments in Clinical IVR Systemsis the ability to manage delivery of trial supplies and studymedications. While the ability to request delivery of theseitems via IVR has been around for some time, the manage-ment of the delivery is causing some sponsors to take a closerlook.

Traditionally, the sponsor sends investigational sites pack-ets, or kits, for trial subjects. Each kit is outfitted with all ofthe supplies and materials, sometimes including study medi-cation, for the duration of the trial. For lengthy trials, trialcosts can increase dramatically when subjects withdrawearly and the content of the kit is wasted, especially whenkeeping subjects enrolled is difficult, or for drugs that arevery costly to manufacture.

To remedy this situation, and potentially save hundreds ofthousands of dollars in the process, some Clinical IVR Sys-tems are providing “just-in-time delivery.” This is a coopera-tive effort between the IVR vendor and the sponsor. The IVRsystem requires the sites to enter study participation data oneach subject, and the sponsor breaks up kits into separatedeliverables for each visit.

After each visit, the sites make a very quick call to the IVRsystem and record the visit in the database. After verifyingthe subject is still in the study, a report is sent to the drugshipment facility, and/or the trial supplies facility, and mate-rials are shipped to the site for the subject’s next visit. If asubject is lost to follow-up, or withdraws consent, etc., thesponsor saves the cost of the drug, the materials, and also the

Page 35: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Clinical IVR Systems

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

shipping. The cost savings, in addition to the minimal addi-tional effort required, is more than enough to make such drugand supplies management a necessity for all clinical trials.

Potential DrawbacksIn the aforementioned scenario of granting waivers, whichactually happened repeatedly to a real-life trial during Visit2, the IVR help desk would receive a call from the Investigatorsaying, “the subject is in the waiting room, and the IVRsystem says the subject does not qualify to be randomized.What do I do?” Unfortunately, the IVR help desk could notresolve this particular problem because it required a clinicaldecision. A member from the clinical team had to review thedata and make a decision to withdraw or randomize thesubject, which could potentially lead to a high level of frustra-tion for the investigational sites.

Such scenario also reflected negatively on the IVR systemitself, even though the IVR system was doing its job perfectly.To prevent this from happening, a careful evaluation at thebusiness processes is crucial during the requirement phase ofthe IVR system so that developers can implement propercontrols and functionalities to handle anticipated scenarios.For example, the frustration could have been eliminated byadding a web portal where investigators can make clinicaldecisions on-line or by sending them automated alerts whenthe IVR system detects a low randomization rate.

It is a potential downside that using “off the shelf” solutions;that is, most of them require you to do things their way, whichcan be quite peculiar sometimes. In addition, some of thesesolutions may not be specialized for the biopharmaceuticalindustry. For any IVR system with regulatory impact, you canexpect to add functionality and documentation to make themcompliant with FDA 21 CFR Part 11.

ConclusionIVR systems for the biopharmaceutical industry are stillgrowing in size, complexity, and functionality. The number ofcompanies taking advantage of Clinical IVR Systems isslowly growing, but is still far from a majority. However,there are a variety of clinical applications for IVR systems,such as those related to neuropsychopharmacology and psy-chiatry.

More people should be aware of the capabilities, timesavings, and potential cost savings of Clinical IVR. Consider-ing the myriad of tasks that can now be accomplished over thetelephone, integrating an IVR system to a clinical trial willcertainly allow biopharmaceutical companies to produce ro-bust and reliable results.

References1. Wave files are a Windows standard sound file format.

Many, but not all, telephony cards support one or moreWave file types like mono, stereo, etc.

2. Refers to Dialogic VOX, a proprietary voice file formatusing adaptive differential pulse-code modulation(ADPCM)

3. Paper diaries rarely translate into useable IVR scriptswithout some kind of alteration. For example, diaries thatrequire date and time entries have much greater successwhen the diary script is changed to an event-based format.The IVR system, not the subject, then records the time anddate of each event.

About the AuthorAnthony Chow recently joined OctagonResearch Solutions’ Clinical IT Departmentto research and develop IT solutions for alllines of business. His work at Octagon hasbeen related to Electronic Data Capture(EDC) using Interactive Voice Response (IVR)systems. He also is responsible for develop-ing CDISC-compliant solutions for regula-

tory submissions. Prior to joining Octagon, he was a seniorconsultant with BearingPoint, a global IT consulting com-pany formerly known as KPMG Consulting, where he heldleadership positions at several high-profile projects withFortune 50 companies and government agencies. Chow holdsa BA in computer science from Northeastern University. Hecan be contacted by tel: 1-610/265-8300 or by e-mail:[email protected].

Octagon Research Solutions, 1016 W. 9th Ave., Suite 102,King of Prussia, PA 19406.

Page 36: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

Renovation Versus New Buildby Werner Seiferlein

This articlecompares thebenefits andpotentialdifficultiesinvolved ineitherrenovating anexisting facilityor building anew facility. Introduction

The decision to improve a plant may beprompted by changes either in internal(technology/business/company) or ex-ternal (regulatory/governmental/envi-

ronmental/political) environments. Choosingbetween renovating an existing facility or build-ing a new one is rarely simple and is frequentlybased on a balance of many factors, such astotal cost of ownership, production require-ments, and regulatory aspects, all of whichrequire thorough consideration.

Reasons for Improving aPharmaceutical Plant

The reasons behind a decision to improve anexisting plant may be both external and inter-nal. External reasons, initiated by authoritiesor market forces outside the company, mayrequire that a facility be modernized. Changesto regulatory requirements, local Environmen-tal, Health and Safety (EHS) requirements, orenergy saving may require that a facility be

enhanced to maintain compliance with thoserequirements. A facility that is not in compli-ance with current GMP requirements may re-quire Renovation to realize compliant status.

There are several reasons from within anorganization that may prompt an improvementin a plant. If the production performance of theplant deteriorates (or plateaus), a businessdecision may be taken to invest in/improve thatplant to increase performances in production.It may be that the company wants to expand orreduce production of a product which requiressome form of modernization.

Maintenance of the plant also may be adeciding factor. Where production units are notreliable, Operational Equipment Effectiveness(OEE) deteriorates, and may indicate the incor-poration of alternative machinery. Unreliableunits may no longer be reparable, or are simplytoo costly to repair and require replacement. Itmay be that the company decides that it is timeto introduce innovative technology to boosttheir production and cost effectiveness and the

Figure 1. Increasinginnovation by deciding aNew Build.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 37: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 2. Total cost of ownership over time.

existing facility design may restrict this policy.Political or strategic decisions within a company also may

determine improvements to a plant. The local tax conditionsmay change, or the company’s relation to a particular countryor market may modify. The company ‘vision’ and strategymay alter, particularly where there has been a merger. Thecapital expenditure situation, cash flow, and market forcesmay require that the time to market for a particular productbe altered.

All such decisions may prompt a determination of whetherthe plant needs some form of improvement. Once this isdecided, the question of “Renovation versus New Build” hasto be considered.

The outcome of a 2003 study1 showed that approximately90% of investment projects are in the “Renovation” category.In contrast, approximately 40% of the investment was spentfor the remaining 10% of ‘New Builds.’

User Requirements Follow Site StrategicThe success of a project2 depends on several factors.3,4 One ofthese factors is to know the goal and the requirement of aproject. Applied to “Renovation versus New Build,” answersto the following questions should provide clear goals andclarity for future steps:

• How might business change over time?• What is the required product volume and capacity?• How much space will be needed?• What kind of influence is infrastructure likely to have?• How will the space support the technology?• What are the future personnel needs?• What are the tax liabilities likely to include?• Will depreciation limitations prevent efficient renewal?

Building - LayersThe concept of “Renovation or New Build” needs to considerthe different levels of that Renovation or “New Build.” Oneway to achieve this could be to look at a building as a six-layered system with each layer having distinct implicationsin time.5

The construction sequence or six-layered system follows astrict order:

“Site preparation, then foundation and framing thestructure, followed by skin to keep out the weather,installation of services, and finally space plan or layout.Then the tenants truck in their stuff.”

The six layers are indicated by:

• skin• structure• services• space plan (or layout)• stuff• site

The exterior surface, known as ‘skin’, has to be changed every20 years, in accordance to aesthetic technology or mainte-nance needs. ‘Structure’ denotes the foundation and load-bearing elements with a lifetime of 30 to 300 years, which areexpensive to change. The interior elements of a building, suchas communication cabling, electrical wiring, and HVAC areknown as ‘services’. These have a lifetime of about 15 years.The fourth layer is named ‘space plan’ and includes theinterior elements such as walls, ceilings, and doors. The ‘stuff’(trucked in by the tenants) includes furnishings. The lastlayer is ‘site’; compared with the other layers, this layer hasprobably the most timeless and enduring attributes.

As one looks at reclaiming older buildings for modern use,in most cases, three or four of the six layers have to be alteredor dramatically revamped. Engineers and architects mustreach beyond the skin, the space plan, and deep into theservices layer to fully update and reclaim outdated buildings.

How Space Can Affect WorkAll six attributes combined create the efficiency of a building.The space exists only as the forum in which the pharmaceu-tical processes and infrastructure are performed. However,space can have a tremendous impact on the quality, function,and speed of the work. It also can have a significant impact onthe costs of supporting the work performed.

Building design can influence new ways of working, sup-port user control and comfort, and supports the functionalityof technology.

RenovationThe Renovation of existing buildings can yield flexibility,timing, and cost advantages over new construction.

In How Buildings Learn, author Stewart Brand5 arguesthat buildings shape, and are shaped by, both space and time.Intelligent Renovation is as much an art as original design.Antiquated services, environmental hazards, and inadequateaccessibility often plague older buildings. Poor lighting, in-sufficient cabling infrastructure, and inadequate Heating,Ventilation, Air Conditioning (HVAC) systems are commonafflictions in such structures, even those less than 10 years ofage. While the structural grids of some existing buildingsmay prevent cost-effective Renovation, many “obsolete” build-ings can be successfully upgraded for contemporary use.

Page 38: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

Renovation may afford certain advantages over the alter-native of a New Build. Use of an existing site or a new‘greenfield’ site both have distinct pros and cons, such as timeout of production during demolition and ‘New Build,’ oracquiring and training of staff.

Renovation of an existing facility may be the faster option,improving the time to the market for any products affected bythe decision to renovate or rebuild, and present less risk torealize the extended product manufacturing. For a Renova-tion the infrastructure is already in place:

• energy (e.g., waste, water, gas, electricity)• utilities (e.g., power-gas supply, steam, IS)• transport connection (e.g., road, rail, air)• real estate (e.g., climate, ground-water)• adjacent residential areas• legal position and development regulations

During a Renovation, staff is already present and any exten-sion to their training is likely to be less time consuming.The financial feasibility of Renovation depends on the exist-ing investment in a building. Depreciation schedules, debtratios, operating costs, revenues, and tax liabilities are allpart of the equation.

Contrary to the decision to renovate is the need to freesufficient space to realize the Renovation (interim manufac-turing, space area) along with constraints in terms of layoutand building design, and the possible limitations in applying

new technology. Renovation may need to continue duringrunning production or manufacturing and the potential con-sequences need consideration. Existing facilities may need tobe demolished to make room for the Renovation in advance,and this could impact current production. A step-by-stepRenovation under ongoing production requires adequate substeps or even provisional relocation of manufacturing whichcould extend the required timeline. Alternatively, establish-ing a stock of product can bridge the time of Renovation.Another solution could be to relay prefabricated modules,which can “push” in the plant where this is possible.

New BuildNew Builds allow for a quantum leap within technology (andknowledge). Such building allows for the creation of futureorientated building design, in which a long-term view may beaccommodated. The entire build may be performed accordingto current EHS ergonomic and GMP requirements. Studieshave shown that ultimately this option will lead to a lowerTotal Cost of Ownership (TCO).

With continuous Renovation and adaptation to new tech-nology, the limits of improvement eventually will be reached- Figure 1. However, if planning and design take account ofavailable and appropriate new technology, it is possible to‘achieve’ the so-called ‘quantum leap’ in innovation - Figure 1.Innovation in this regard is a process involving multipleactivities, performed by multiple persons from one or severalorganizations, who develop an extraordinary and extreme

Figure 3. The evaluation of workload and time during the project phases for Renovation and New Build.

Page 39: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 4. Original layout of the packaging area.

new process, equipment, and/or technology.6

The implementation of new technologies may carry a highrisk because these technologies are not proven by regularbusiness use and may not provide the necessary function orreliability, so this needs to be assessed and defined. Time andresources are also factors to be considered and may countagainst a New Build. It may take some time to locate andacquire an appropriate site, and staff will need to be hired andtrained.

It may be cheaper to have a New Build, rather than toinvest in “stop gap” solutions, but a New Build also requiresa high immediate investment.

Cost and TimeIn order to ascertain whether Renovation or New Build is thebest way forward, innovative technology business drivers,such as cost and time, need consideration.

CostWhen referring to cost, the complete picture of cost, i.e., theTotal Cost of Ownership (TCO), should be considered.

The total cost of a facility is the sum of:

• one time capital cost (initial cost)• initial cost includes the purchase price, first set of spare

parts, plus engineering, installation, commissioning, vali-dation, and training

• operating costs of the facility throughout its lifetime

• operating costs may include raw materials, energy andother utilities, manpower, and environmental factors

• maintenance costs of the facility throughout its lifetime• maintenance costs are cost of Predictive and Preventive

maintenance, plus cost of repairs• other operating costs where a failure of the facility leads to

plant shutdown• failure costs include wasted raw materials cost, off-spec

cost, clean-up cost, disposal cost, profit loss, etc. resultingfrom non-availability of the facility

• dismantling/disposal costs or residual value• dismantling/disposal cost is the cost to get rid of the

facility, clean the area, and make it available for the nextuser. If the facility can be sold, it is a value.

The goal must be to choose facilities which minimizes thelifetime cost of ownership; this minimum is often a balancebetween initial cost of the plant and lifetime operation andmaintenance costs. This implies that the project managershould utilize the know-how and expertise of various func-tions, such as engineering, purchasing, operations, qualitycontrol and assurance, process development, as well as ofplant personnel and suppliers.

In all cases, a Net Present Value (NPV) evaluation shouldbe made. The other indicators must be applied independentlyof the type of project and the decision that has to be supportedby the evaluation.

Page 40: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

For the NPV calculation, any available system may beused, and the methodology presented in this article applied.

Value engineering should be performed in order to opti-mize projects with a structured methodology in addition tocurrent project best practices. This is a structured method tooptimize the project’s value, starting with the challenge ofproject objectives, assuring consistency between needs,functionalities, and technical solutions with minimum TCOor maximum NPV as the target.

At some point in the lifecycle of a pharmaceutical plant,the cost of Renovation exceeds that of a New Build. AlthoughRenovation may seem a less costly option initially, overall itmay be more expensive than building a new plant - Figure 2.

In the case of Renovation (Figure 2), investment is step bystep in relation to requirements, e.g., compliance or otherrequirements of plant improvements. The total cost of owner-ship will increase in regard to time.

In the case of a New Build (Figure 2), there is a majorinvestment at the beginning of the project, but subsequentlymore efficient equipment processes and appropriate layouts,which need less space. The total cost of ownership will belower.

The white arrow (Figure 2) shows the break-even-pointwhen it is possible to take advantage of the New Buildscenario. The time when the break-even-point occurs de-pends on specific circumstances.

TimeTo answer whether the Renovation of a plant or a New Buildwould take more time, the specific parameters of a project areneeded. This method is an attempt to undertake a compari-son of the relationship between project phases.

A project consists of several phases.7 In the case described,six phases are defined - Figure 3:

1. feasibility study with the key activities:• define project scope, project boundaries, regulatory

constraints, etc.

2. project development with the following key activities:• start to set up user requirements, preliminary cost

estimation, and time schedules, consider alternativesin conceptual design, etc.

3. basic engineering with the following key activities:• basic drawings, cost estimation for capital authoriza-

tion request. This is the step where total cost of owner-ship and value engineering should be applied andinvestigated.

4. detail engineering with the following key activities:• final drawings and document specification, etc.

Figure 5. Renovated layout of the packaging area - adapted concept to requirements.

Page 41: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

6 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Renovation Pros• Faster in Realization• Less Risk to realize the extended production manufacturing• Infrastructure is given

- Energy (Waste, water, gas, current, etc.)- Utilities (Power-gas supply, Steam, IS, etc.)- Transport connection (Road, Rail, Air, etc.)- Real estate (Climate, Groundwater, etc.)- Adjacent residential areas- Legal position and development regulations

• Staff is available

Table A. Pros and cons of Renovation.

Renovation Cons• Demolishing of existing facilities and equipment in advance• Need sufficient free space to realize Renovation (interim manufacturing,

space area)• Renovation during running production or manufacturing• Constraints in terms of layout and building design• Limited in applying of new Technology

5. construction with the following key activities:• final acceptance of construction work, execute IQ and

OQ protocols, etc.

6. commissioning and qualification with the following keyactivities:• conduct pre start-up checks, execute qualification pro-

tocols, release to operation, etc.

Figure 3 shows the evaluation of workload and time duringthe project phases for Renovation and New Build.

In the first phase, Feasibility Study Renovation and NewBuild are likely to have the same time and workload require-ments.

In Phases 2, 3, and 4, you should need more workload andtime for New Build to establish documents like drawings,layouts, and specifications for equipment and devices. In thecase of Renovation, existing documents may be utilized.

In the case of construction, more time is required forRenovation, as production needs must be met, so construc-tion work continues in parallel to production and a kind ofinterim manufacturing space.

Hence, both Renovation and New Build have factors whichcause increases in the time needed.

In case of Phase 6, “Commissioning and Qualification,” thesame argument given for Phases 3-5 could be applied; exist-ing documentation and work may be relied upon.

This is a very rudimentary judgment, as different timeperiods are required for each project phase. Nevertheless, bycounting the “+” and “-” it seems that Renovation could befaster than a New Build.

Example for RenovationPackaging Area RenovationA packaging area required a new layout to facilitate a changein company philosophy. This change required that the openhandling of non-sterile pharmaceuticals in the packagingarea must be performed in a separately controlled environ-

mental zone. The decision to renovate rather than rebuildthis packaging area caused several difficulties, which re-quired specific solutions, while minimizing disruption toproduction. Figure 4 shows the original layout of the packag-ing area. Figure 5 shows the renovated layout of the packag-ing area.

The Renovation could be achieved by segregation of pri-mary and secondary packaging or by local protection of thefilling section of the primary packaging machine. Difficultiesin implementing local protection on existing packaging ma-chinery compelled the segregation of primary and secondarypackaging areas.

This required the packaging area to be divided into twodistinct environmental zones, classified as E1 and E2, whereE1 was the more stringently controlled environment for theopen handling of non-sterile pharmaceuticals. Consequently,the gowning areas required segregation and separate HVACsystems had to be constructed to provide for the two distinctenvironments.

Several difficulties arose with the decision to renovate,including:

• finding an acceptable location for the segregation wallbetween primary and secondary packaging

• implementing a separate personnel access route for thesecondary packaging area

• rearranging the entire HVAC system• realization without unacceptable manufacturing inter-

ruptions• reorganization of the operation and maintenance activi-

ties, required by the segregation

The example shows a remodeling project initiated by compli-ance issues, the solution to which demonstrates the difficul-ties that arose, and the compromises necessary to remedythose difficulties.

The initial investment required may be considered as theinitial step - shown on the red curve in Figure 2. A single

New Build Pros• Quantum leap within Technology and Knowledge• Create future orientated building design• According to current EHS, ergonomic and GMP requirements• Less Total Cost of ownership

Table B. Pros and cons of New Build.

New Build Cons• High Risk with implementation of New Technologies• Probably needs time for Site Selection• Hire and train new staff

Cheaper than to invest in “stop gap solutions” but High investment at once

Page 42: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Renovation versus New Build

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 7©Copyright ISPE 2004

investment such as this is unlikely to justify a New Build.However, where further investment is required (as signifiedby the stepwise investments indicated in Figure 2), anyinvestment made after the position indicated by the whitearrow could be considered squandered.

Since the investment for Renovation occurs step-by-stepover a long timeframe, it may be difficult to predict futureinvestment. A better approach could be to establish a sitestrategy to define and identify the potential triggers andrequirements that lead to Renovation. This allows the calcu-lation of the cost of these investments for a clear period oftime, and a comparison to the cost of a New Build to be made.

ConclusionsRenovation of existing areas seems to be faster comparedwith a New Build - Tables A and B.

This occurs because during the project phases ProjectDevelopment, Basic-and Detail-Engineering, and Commis-sioning and Qualification, a significantly higher workloadand more time is needed to establish drawings, layouts, andspecifications for equipment and devices.

Despite this consideration, a general decision to selectbetween the Renovation of existing space or a New Buildspace is not feasible.

The decision strongly relies on the conditions and circum-stances related to the specific goal and content of a project.

Existing tools and methods, such as Site Master Planning,Total Cost of Ownership, and Value Engineering which coverwell-known calculations, e.g., Net Present Values (NPV) canhelp support such decisions.

Decisions for a New Build depend on economical reasons,but also may require top management to have the courage oftheir convictions to initiate such projects.

References1. Seiferlein, W., Analysis and Improvement of Planning-

Quality of Investment-Projects (Study), 2003.

2. A Guide to the Project Management Body of Knowledge,(PMBOK® Guide), 2000 Edition, Project Management In-stitute (PMI), ISBN: 1880410230.

3. Slevin, D.P. and Pinto, J.K., The Project ImplementationProfile, New Tool for Project Managers, Project Manage-ment Journal, September 1986, PP 57-70.

4. Lechler, T., Erfolgsfaktoren des Projektmanagement, Pe-ter Lang, Frankfurt am Main, 1997.

5. Brand, Stewart, How Buildings Learn: What HappensAfter They’re Built, Penquin Books, June 1, 1994 ISBN:0670835153.

6. Gemünden, H.G., Management of Innovation I, Chapter 1,The Object of Analysis: “Innovation,” 2003.

7. Craigmile, Mitchell G., “Applying Project ManagementTechniques to Achieve Facilities Certification,” Pharma-ceutical Engineering, Vol. 13, No. 1, January/February1993, PP 16-24.

About the AuthorWerner Seiferlein is Director IndustrialEngineering at Aventis, Sterile Operationslocated in Frankfurt. He studied mechanicalengineering at the Technical University ofDarmstadt and holds a masters degree inmechanical engineering. Seiferlein has morethan 19 years of experience in the field ofActive Pharmaceutical Ingredients and Drug

Product Manufacturing, as well as Facility Management. Hemanaged various remodeling and new investment projects inGermany as well as abroad and has obtained essentialknowledge in the fields of process analysis and optimization.He also was responsible for a selection of the Hoechst MarionRoussel Headquarters and the move for about 500 workplaces. Before his current job, he was Head of Engineering forProcess Development and R&D of Aventis Pharma GmbH,Germany and Head of DP Global Engineering of Aventis.Seiferlein is a member of ISPE’s Technical Documents Facili-ties Subcommittee. He can be contacted by e-mail:[email protected].

Aventis Pharma, Aventis Pharma Deutschland GmbH,Building H 650, 65926 Frankfurt/Main, Germany.

Page 43: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Isolator Campaigning

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 1©Copyright ISPE 2004

Isolator Campaigning - An IndustrySurvey and Discussion of OperationalPracticesby Douglas Stockdale

This articlediscusses therecent trend ofCampaigningsterile drugbatches inconjunction withthe installationof isolatortechnology foraseptic fill/finishoperations. Thearticle wasbased on apresentationthat was held atthe 11th AnnualBarrier IsolationTechnologyForum inArlington, VA,June 2002.

Figure 1. Campaigntrends.

Introduction

Campaigning, the manufacturing of mul-tiple batches of parenteral fill/finishproducts between complete sanitiza-tion/sterilization processes may be an

economic necessity with the implementation ofisolator technology for high speed aseptic fill-ing operations. This was one of the key pointsthat came out of a recent independent isolatortechnology survey.1

The purpose of the Isolator Campaign sur-vey was to provide the industry with opera-tional and compliance information regardingthe implementation of a Campaign process inconjunction with Isolator Technology. This wasa subject that Mike Winter, Director Form-Finish, Baxter Biosciences, and I, President,Stockdale Associates, Inc., had personally noted

that kept reoccurring in a number of side dis-cussions at ISPE Isolator Technology confer-ences in the late 1990s. At the 2001 ISPEBarrier Isolation Conference, “Isolator Cam-paigning” was a largely attended break-outdiscussion group meeting. Subsequently, MikeWinter, and I initiated an independent indus-try survey on the use of Campaigning withinthe Isolator user community. Jack Lysfjord,Vice President, Valicare division of Bosch Pack-aging and Co-Chair of the ISPE Barrier Isola-tion Conference, provided us with valuableguidance during the entire survey process.

The isolator technology community is in atransitional period. It has overcome the techni-cal development hurdles of the Isolator processas applied to the Life Science Industry. Mosthave come to recognize the ability of isolator

technology to further reducethe risk of microbial con-tamination of asepticallyfilled products. Isolator filllines are now facing the in-tense and tough operationaleconomic requirements.

The high-speed, largeproduction volume isolatorfill-lines are faced with thecommercial reality of pro-duction schedules, through-put, return on managed as-sets, and standard cost. Abig concern with the imple-mentation of isolator tech-nology is the extended du-rations required for the sani-tization cycle of an isolatorwith Vaporized HydrogenPeroxide, which is becom-ing the industry de facto

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 44: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Isolator Campaigning

2 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Figure 2. Campaign durations – current.

standard. One way that organizations could possibly lever-age their isolator investment is to utilize the process ofCampaigning, which is to produce multiple batches of prod-uct between sanitization/cleaning cycles.

The Isolator Campaigning survey was developed to under-stand the current industry thinking about the implementa-tion of Campaigning in conjunction with isolator technology.The survey results were expected to be used to benchmark theindustry on how the use of Campaigning could impact thecontinuing implementation of Isolator Technology.

The concept of Campaigning was already being consideredas an operational tactic with Isolator Technology by a numberof early innovator companies. Many pharmaceutical compa-nies had a history with Campaigning for batches of ActivePharmaceutical Ingredients (APIs) produced between clean-ing cycles. Biotechnology companies also were gaining someoperational experience with Campaigning when the conceptis applied to the use of chromatography columns.

The IsolatorCampaigning Survey

In a preliminary meeting about this survey, it was decidedthat Campaigning would be really relevant to large-volume,commercial users of isolator technology. Small scale isolatorusers would probably not be under the same scheduling andthroughput constraints, therefore Campaigning would prob-ably not be an issue that would need to be addressed. It wasrecognized that the resulting survey would then be limited toa very small and restricted distribution (22 companies sur-

vey), but would be a very focused and relevant survey.The importance of keeping this survey confidential and

reporting the results anonymously also was recognized as theissue of how to implement a Campaign process is becoming asensitive issue. The survey respondents had an opportunityto return their survey to either Winter (Baxter) or me (Con-sultant). This concern did prove true, as we received two non-responses due to “management objections to discuss a criticaloperational strategy.” Otherwise, a very high response rate ofmore than 80% was received (22 surveys distributed, fournon-response); which reflects the industry’s desire to shareinformation and further improve the safety of patients whowill be utilizing the sterile products. The data and graphspresented in this article will not add up to a full 100% as thesurvey data is effected by partial survey responses, wheresome questions have purposefully omitted answers, andwhere there were multiple isolator installations, with opera-tional variance between geographies and installations. Wesincerely appreciate all the time that was spent by all of thosewho invested their time and shared their information withus.

The interim study report was presented at the 2001 ISPEEuropean Isolator Technology Conference in Zurich. Thefinal survey results were presented at the 11th Annual BarrierIsolation Technology Forum in Arlington, VA, June 2002.

Survey ResultsFirst, more than 70% of the operating companies, which arenow considering an investment in high-speed isolator tech-

Page 45: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Isolator Campaigning

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 3©Copyright ISPE 2004

nology, also are evaluating a Campaign process – Figure 1.When isolator technology was being initially developed, lessthan 20% of the companies had considered the use of Cam-paigning as an operational option. Comments were receivedthat early high-speed Isolator lines did not have many of thecurrent options available that would easily facilitate a Cam-paign process.

All the comments about reasons for implementing isola-tors were related to the increased ability to control microbialcontamination by eliminating personnel. The principal rea-son to implement a Campaign process was economics: re-duced down time. Companies commented on the increasedloss of production time with isolators for sanitization be-tween batches versus their traditional barrier aseptic filllines. Without Campaigning, they were not getting the samevolume of production per fill line, which was compounded bythe increased investment for the isolator technology, thusincreasing the manufacturing standard cost per vial. Toimprove their economics, e.g. reduce their manufacturingstandard cost per vial, they needed to implement Campaign-ing and reduce downtime.

A comment from a number of organizations, which havealready made an isolator technology investment, is the addi-tional investment required to modify their equipment topermit Campaigning. Others may not consider the imple-mentation of Campaigning now, they would consider Cam-paigning with future isolator technology investments.

Of those companies that are not considering Campaign-ing, half cited United States Food and Drug Administration(FDA) regulatory concerns. The others either produced verylarge batches now and Campaigning would not provide anadvantage or there were other product issues that wouldpreclude Campaigning.

Second, there was an initial large difference in the use ofisolator Campaigning for products manufactured and dis-tributed in the European Union (EU) versus North America(NA)/United States. At the time of this survey, there is nowmore of a balance in Campaigning between the two geogra-phies – Figure 2. The majority of the European isolatororganizations are Campaigning products exclusively for Eu-ropean distribution. The European isolator organizationsstated that they are cautious about Campaigning productsintended for export and commercial distribution within theUnited States. Many of the United States/North Americanisolators are not Campaigning, but were very interested inthe process and considering the potential implementation.Every comment regarding the implementation of Campaign-ing in the United States/North America was directed at theuncertainty of the FDA’s position on Campaigning, the re-quirements that would need to be met to validate a Cam-paign, and the potential of regulatory approval.

Third, organizations focused on the duration of the Cam-paign, versus the number of batches processed, but recog-nized there was a finite limit of batches that could be pro-cessed in that duration. There is no consensus on the numberof batches or durations of a Campaign process – Figure 2. Butthere is a consensus that the number of batches or the

duration of the Campaign could be greater than was initiallythought, and more than 70% of those Campaigning areconsidering a longer Campaign duration. Figure 3 highlightsthe responses to the survey that provided target durations forextending their Campaigns.

The comments about Campaign duration indicated thatdecisions were determined by either compliance/operational/finance risk about by a potential “failure” or the amount ofrevalidation time required to support an extended duration.Most organizations did not want to comment about failuremodes at this time.

Fourth, there was a 100% consensus by all organizationsthat only products of the same family would be Campaignedwith variations occurring only in potency within the sameproduct family.

Fifth, all of the organizations were requiring the use ofClean-In-Place (CIP) and Steam-In-Place (SIP) on the fluidpath of the product between batches during a Campaign.

Sixth, there was a consensus (80% of those who answeredthis question) that the media fill would simulate the maxi-mum lot size of the Campaign – Figure 4. Most of theorganizations indicated that the media fill for validation wasa concern, and very few organizations provided detailedinformation about their media fill strategy.

DiscussionEarly high-speed isolator lines were not designed to easilyfacilitate a Campaign process, such as the inclusion of CIP/SIP for the product fluid path. The possible reason for notconsidering Campaigning in conjunction with early isolatortechnology development was two fold: aseptic filling opera-tions are traditionally a single batch process as dictated byregulatory guidance, and second, the amount of clock timethat was going to be required for Vaporized Hydrogen Perox-ide sanitization was not well understood.

The commercial implementation of isolator technologywas adopted quicker in the EU, and likewise the use ofCampaign processing also is being quickly adopted for isola-tor technology by EU, but not for products to be commerciallydistributed in the United States.

Even though there is stated FDA support for the imple-mentation of isolator technology as an improvement on asep-tic fill/finish, most organizations are cautious about seekingFDA approval for Campaigning of aseptic production in

Figure 3. Campaign durations – potential.

Page 46: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Isolator Campaigning

4 PHARMACEUTICAL ENGINEERING SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

conjunction with isolators.Prior to our presentation in June 2002, an extensive

literature search was conducted of the FDA guidance docu-ments regarding Campaigning. The only Campaign guidancethat we presented then, as it is now, is within the Interna-tional Conference on Harmonisation (ICH) Q7A, “Guidanceof Industry; Good Manufacturing Practice Guidance for Ac-tive Pharmaceutical Ingredients,” dated August 2001. ICHQ7, Section 5.2, Equipment Maintenance and Cleaning states:

“When equipment is assigned to continuous production orCampaign production of successive batches of the sameintermediate or API, equipment should be cleaned atappropriate intervals to prevent build-up and carry-over contaminants (e.g., degradants or objectionablelevels of microorganisms).”

A problem for the Campaigning practice is that the scope(section 1.3) of the Q7A, states: “The sterilization and asepticprocessing of sterile APIs are not covered by this guidance,but shall be performed in accordance with Good Manufactur-ing Practices (GMP) guidance for drug (medicinal) productsas defined by local authorities.” Even so, during the Ques-tions and Answer discussion following the June 2002 presen-tation,1 Richard Friedman, FDA, stated that Q7A did helpestablish guidance for Campaigning with isolator technol-ogy. It was still necessary for an operating company tovalidate what would be an “appropriate interval” for a Cam-paign production.

A review of the Draft Guidance for Industry, Sterile DrugProducts Produced by Aseptic Processing - Current GoodManufacturing Practice, August 2003, does make two state-ments that address the process of Campaigning of Asepticfilled drugs.

First, Section IX, Validation of Aseptic Processing andSterilization, Section C (line 042) states “Sterility of asepticprocessing equipment should be maintained by batch-by-batch sterilization.” The challenge to the operating companyis how to work within the language of this statement toperform a Campaign process.

Second, in Appendix 1, Aseptic Processing Isolators, sec-tion D3, Decontamination Frequency, states: “When an isola-tor is used for multiple days between decontamination cycles,the frequency adopted should include a built-in safety marginand be well justified.” This statement does not provide the fullresolution with isolator technology processes for the batch-by-batch statement in Section C, above.

It also could be argued that this draft document does notexplicitly prohibit Campaigning of multiple batches either.Reference is made in Section E, Design, to 21 CFR 211.67(a)that “Equipment and utensils shall be cleaned, maintained,and sanitized at appropriate intervals to prevent malfunc-tions or contamination that would alter the safety, identity,strength, quality or purity of the drug beyond the official orother established requirements.” The use of the term “appro-priate intervals” is the same terminology as noted in Q7Asection 5.2 and could permit an operating company to vali-date what an appropriate interval is for their productionprocess, which includes the process of Campaigning.

Section VIII, Time Limitations, of the Draft Guidancerequires only that “time limits should be established for eachphase of aseptic processing,” which does not prohibit thevalidation of multiple batches as Campaign production pro-cess.

Some Campaign validation guidance is provided in Sec-tion IX, Validation of Aseptic Processing and Sterilization,section A.1 Size of Runs. The Draft guidance states: “Somebatches are produced over multiple shifts or yield an unusu-ally large number of units, and media fill size and durationsare especially important considerations in the media fillprotocol. These factors should be carefully considered whendesigning the simulation to adequately encompass condi-tions and any potential risks associated with the largeroperation.”

One issue that may concern the FDA is the challenge to thelong tradition of single batches of aseptic filled products andthe potential Pandora’s box that may now open with theapproval of Campaigning for isolator operations.

If Campaigning is a viable compliance practice with isola-tor technology, why could it not also be applicable to a wellcontrolled “Restricted Access” aseptic operation? Or couldCampaigning be a validated and acceptable process for a verywell controlled traditional barrier aseptic filling operation?

Or extend the application of Campaigning to the sterilelyophilization process, even without the use of isolator tech-nology. Couldn’t a sterile lyophilizer be considered a re-stricted access process if all of the production operations werecompleted through its “pizza” door? The pending applicationof a Campaign process to a tradition sterile lyophilizationprocess was just confirmed during a discussion with a Euro-pean aseptic manufacturing manager at a recent sterileFigure 4. Campaign media qualifications.

Page 47: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Isolator Campaigning

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING 5©Copyright ISPE 2004

lyophilization conference in Brussels.There are no explicit regulatory guidance’s being provided

to assist an operating company with developing and validat-ing a Campaign process for an aseptic filled drug. It would behelpful if industry organizations and operating companiesprovided a consensus input to the isolator section of the DraftGuidance for Industry, Sterile Drug Products Produced byAseptic Processing – Current Good Manufacturing Practiceregarding the validation of Campaign operations.

The industry has now embraced isolator technology andthe process of aseptic Campaigning is both feasible and areality.

References1. Stockdale, Douglas and Winter, Michael, “Isolator Cam-

paigning: ISPE Survey” presentation at ISPE 11th AnnualBarrier Isolator Technology Forum, June 5-6, Arlington,Virginia 2002.

2. Draft Guidance for Industry, Sterile Drug Products Pro-duced by Aseptic Processing - Current Good Manufactur-ing Practice, August 2003.

3. Q7A Good Manufacturing Practice Guidance for ActivePharmaceutical Ingredients, September 2001.

4. Sterile Drug Products Produced by Aseptic Processing,June 1987.

5. Parenteral Drug Association (PDA) Technical Report No.34, Design and Validation of Isolator Systems for theManufacturing and Testing of Health Care Products,published September 2001.

6. PDA Technical Report No. 36, Current Practices in theValidation of Aseptic Processing - 2001, published May2002.

7. PDA Points to Consider for Aseptic Processing, published2003.

About the AuthorDouglas Stockdale is the President ofStockdale Associates, Inc., which providesextensive aseptic fill/finish and sterile pack-aging consulting services for the Life Sci-ences industry. He had 20 years of opera-tional experience with Baxter Healthcareprior to founding Stockdale Associates. He isan internationally known consultant,

speaker, and writer about the issues of aseptic fill/finish andsterile packaging. His company provides innovative strategicand implementation solutions for product development, manu-facturing, compliance/auditing and validation. He is a mem-ber of PDA, AAPS, IOPP, and ISPE, and a member of theBoard of Directors for the ISPE Greater Los Angles Chapter.Stockdale has a MBA from University of La Verne and BS inengineering from Michigan State University. He can bereached by tel: 1-949/888-9488 or by e-mail:[email protected].

Stockdale Associates Inc., 10 Reata, Rancho SantaMargarita, CA 92688.

Page 48: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Outsourcing Clinical Supplies

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 1©Copyright ISPE 2004

Strategic Value of Clinical Suppliesby Colin Andrews

This articleexplores theneed for a morestrategicapproach tooutsourcingdecisions inclinical materialsmanagement.

Background

A previous article in PharmaceuticalEngineering1 described the dynamicand changing business environmentfor drug discovery and development.

This article concluded that – due to a potentialstep increase in drug discovery – executionprocesses such as Clinical Materials Manage-ment (CMM) are becoming more critical tocompetitiveness than ever before.

A key area of focus in CMM has been aboutthe development of outsourcing in this area ofactivity.2,3 For some firms, this may be the onlyviable option for effective CMM. For others,outsourcing may be appropriate in selectivecases. Indeed, the growth of CROs in the lastthree years2 demonstrates the growing demandfor this service.

However, it is equally clear that the decisionto outsource is frequently taken arbitrarilyrather than for strategic/tactical reasons, forexample,“we haven’t the capacity to do ‘this’,‘now.’” Given the increasingly central positionof CMM in determining a company’s competi-tive capability, too many short-term decisionscan weaken its overall competitive performance.The discussion is too often “We don’t have thecapacity so we’ll have to outsource…” The dis-cussion should be “What are our strategic pri-orities?” Are these reflected in our core compe-tencies and capabilities? What must we do our-selves and what therefore should we outsource?”

The growth of the outsourcing of executionactivities has been a feature of many other

industries. There are lessons to be learned andshared. Equally, there are distinctive featuresof the pharmaceutical development industrythat must be taken into account.

This article develops a framework for under-standing where the firm is deriving strategicvalue from its clinical materials processes inorder to help organizations formulate appro-priate strategies for developing their CMMcapability.

Supply Chain Learning from OtherIndustries

Other industries, such as the electronics indus-try, have followed a development path that isanalogous to the pharmaceutical industry, al-beit with relatively more compressed marketlife cycles. In the early growth stages of theindustry, firms were strong innovators. As noone could provide them with the specialistskills and services they required, they weredeveloped in-house. Many organizations be-came highly vertically integrated.

As the industry developed, more and morespecialist suppliers emerged, or were spun off.Original Equipment Manufacturers (OEMs)began to focus on core competences in areassuch as design, assembly, and marketing. Thespecific competencies developed depended onthe OEMs’ particular competitive strategies.Some organizations pursued a strategy of dif-ferentiation where design and new product in-troduction capabilities were core. Others pur-sued a strategy of lowest-cost producer where

Figure 1. ClinicalMaterials Managementcompetencies.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 49: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Outsourcing Clinical Supplies

2 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

design competencies were about following trends and corecompetencies for competitive advantage were in manufactur-ing and distribution.

Toward the end of the growth phase, the typical electronicsproducer would have a number of design ‘centers of excel-lence.’ Production/execution would be via regionally basedmanufacturing centers focused on assembly operations withcore expertise in managing the supply chain. Relationshipswere established with Tier One suppliers who shared theOEMs global reach. Each Tier One supplier would have anetwork of secondary suppliers in each region for the provi-sion of required materials.

With the market essentially mature in many areas ofelectronics products (VCRs, television, personal stereos, com-puters), the OEMs are stepping back from supply chainoperations. Their competitive area of focus is now on productdesign and marketing.

The automotive industry has followed a similar althoughless extreme pattern. The OEMs still tend to assemble thevehicles, but this is not exclusively their preserve. Majorcomponents and sub-assemblies such as engines, gearboxes,and even body shells are produced by partner organizations.These may be suppliers. They may actually include otherOEMs, for example, sharing engines and gearboxes.

In both these industries, close involvement with manyaspects of the suppliers’ operations is seen as critical tomaintaining and developing competitive advantage for theOEM.

Make or Buy in CMMThe decision to outsource Clinical Supplies can be seen as aclassic ‘make or buy’ issue. For some businesses, such as‘virtual’ corporations or bio-technology start-ups, there maybe no other way to access the necessary expertise. For otherfirms, the complete outsourcing of CMM is a dangerouslysimplistic option that ignores the many linkages betweenCMM and other development activities such as study design,formulation development, and manufacturing process speci-fication. It is a common complaint within CMM organizationsthat the trial’s personnel assume that materials are available‘off the shelf’ to meet any trial requirement at any time.

One of the unique features of clinical materials is that theycome in many forms. Not all of these forms are equally‘valuable.’ By the time a drug development program hasreached Phase III, clinical supplies can include:

• the ‘drug’ being trialed - likely to be in short supply andhighly valuable

• a suitable comparator - possibly difficult to source, but oflesser value than the drug

• a placebo - simple to produce and low in value

Issues of blinding in the study will restrict how differentlythese groups of materials can be managed.

It is also necessary to consider sub-divisions of the wholesupply process. Figure 1 describes a simplified supply chainfor CMM.

Any one of the above supply chain steps can be outsourced.It also is possible that only parts of each step may beoutsourced, for example, bulk production of placebo. Theseare important decisions with implications for business per-formance, and such decisions are not to be taken arbitrarily.

Another dimension of complexity in CMM is the phase ofclinical trial being considered, i.e., Phase I – IV. The differentphases have very different requirements both for the volumesof clinical supplies required, and in terms of the level andtypes of controls required. The clinical phase also has asignificant impact on the nature of the manufacturing capa-bility required.

With this complexity, it can be challenging for firms tomanage CMM processes clearly and consistently. It is un-usual to find a consistent approach across multiple clinicaltrials within a single organization. Clearly, some form ofstrategic reference-point for these activities is essential.

The issue is less of a simple and often (capacity deter-mined) arbitrary choice between in-house operation andoutsourcing. The real issue is where most benefit can bederived from outsourcing and what competencies providegreatest benefit to the company if developed and maintained‘in-house.’

Value in the Clinical Materials Supply ChainIt is generally easy for organizations to see the costs involvedin the supply chain for clinical materials. Equally, there is aninherent logic that any facility within a single company set upto cope with a peak of large Phase III trials will be under-utilized at other times. Similarly, the complexity describedabove suggests that any company aiming to maintain a broadcapability in CMM, for its own competitive advantage, mustretain a significant level of redundancy in its Clinical Mate-rials supply chain.

A focus on costs tends to push any outsourcing activity intoa price sensitive transaction-by-transaction equation. Thereis anecdotal evidence that this is the case within CMM.4

Counter-balancing the positive financial attractions inoutsourcing are concerns about the reliability of supply fromcontractors, and worries over assuring the quality of thosesupplies. These concerns are often based on personalizedexperiences of specific projects and often lack appropriatereview of the causes of ‘wrong’ outcomes. The end result ofthis is a tendency to frequently move suppliers, and to imposesignificant levels of intervention in the outsourced processes.

The costs of outsourced clinical materials supply tend tomake up only a small proportion of a typical Phase III trial’sbudget. While significant differences in cost between suppli-ers may only change the overall budget by a few percent,different levels of performance from the supplier (quality,delivery, responsiveness) may negatively impact the wholestudy significantly. Therefore, there is a need to get back tobasics and consider where the ‘value’ in the supply chainresides.

The following elements are suggested as potential areas ofvalue-add for the development company. In turn, these shouldbegin to form the framework for the ‘make or buy’ decisions

Page 50: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Outsourcing Clinical Supplies

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 3©Copyright ISPE 2004

Figure 2. Added value in CMM.

referred to earlier. Figure 2 describes a ‘virtuous circle’ ofadded value from the elements described below.

Product KnowledgeThroughout the development process, there is ‘learning’ abouthow the ‘end product’ production scales-up. Benefiting fromthis ‘learning’ is an underlying driver for ‘concurrent engi-neering’ in other industries. The objective is that – by tacklingproduction issues during the development stage – new prod-ucts can be introduced to market more quickly and withbetter quality. As medicine becomes more ‘niche’ and person-alized, and the development process more transparent withearlier ‘me-too’ products, well-managed product knowledgewill be essential for sustained competitiveness.

This knowledge is of most value to the eventual producerof the drug product. Where arbitrary decisions are madeabout the production of clinical supplies, the producer of thedrug during trials is not necessarily involved intimatelyenough in the later specifying of production processes. Thismakes it difficult to transfer any learning regarding theunique characteristics of the specific product.

It is a particular feature of pharmaceuticals that it isdifficult to make changes to a production process once it hasbeen validated for approval. Early optimization is not just a‘nice to have,’ it is an essential in the new competitiveenvironment. It is also too important to leave to chance.

In general, the management of knowledge regarding theend product can become critical at key stages in the develop-ment. Typical of the types of problems seen are difficultieswith Methods Transfer. These difficulties can arise betweenthe development organization and the contractor, or justwithin the development organization itself. Such difficultiescan cause significant delay in the program. At their worst,they can jeopardize the whole program by introducing con-cerns over the robustness of the product that may or may notin fact be valid.

Problems with this knowledge management can lie unde-tected and may not be recognized until late in the develop-ment program – for example, if there are queries regardingsubmissions to regulatory authorities.

Supply/Lead-Time FlexibilityIt is a fundamental of clinical trial experience that clinicalmaterials supply is not a ‘critical-path’ activity for studystart. Indeed, it has been identified that fewer than 5% ofclinical studies are delayed by late clinical materials.5

However, there are still significant elements of value inreliably reducing the clinical materials lead-time. Researchat the Tufts Centre for the Study of Drug Development hasidentified that reducing the development lead-time by halfwill reduce total costs by approximately 30%.6 This reductioncomes from a myriad of savings, but even from a restrictedCMM viewpoint, the longer materials sit on the shelf, themore effort is required to maintain and coordinate expirydates, the more resource required to identify and managestability issues, and greater volumes of storage required(significant if refrigerated, or specialist storage is required).

Intangible benefits also can come from reducing supplylead-times. The shorter the lead-time, the greater the oppor-tunity there is to fine-tune details of study protocol. Providedthis is well managed, the end result will be a better study. Anexample of how this may work is allowing the study sites tohave more input, or input closer to the study start, therebyimproving investigator ownership of their tasks and reducingthe number of issues to be addressed during the studyoperation.

Alternative UseActive Pharmaceutical Ingredient (API) is often in very shortsupply during the development process. The ability to changethe trial destination of API can be crucial to bringing the bestproduct to market quickly.

One of the mantras often cited for efficient drug develop-ment is ‘kill early, kill often.’ Benefit will only be gained if theresources that are freed up can be redeployed in a timely andpurposeful manner.

This may be simple in principle. However, in practice, thedelays and potential for mistakes in retrieving a batch ofmaterial from one site and re-releasing and/or transferring toanother can, and does, make this course of action impracticalfor some organizations in some cases. Enhancing product andprocess knowledge is essential to achieving these levels oforganizational competence.

Responsiveness to Trial ResultsClinical trials often require ‘fine tuning’ during their run.Patient recruitment may be markedly different from expecta-tions (country, demographics, quantity etc.). Patient reten-tion may be better or worse than anticipated. Survival ratesmay be higher or lower. Results may show un-expectedoutcomes. Each of these can impact the nature of the clinicalmaterials required, and the mechanisms used to manageexisting supplies.

Extended, complex or poorly thought out supply chainscan make change expensive and time consuming. In the worstcase, CMM activities may influence decisions on the develop-

Page 51: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Outsourcing Clinical Supplies

4 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

ment process. For example, as an extreme case, an otherwisepromising drug program may fall into the ‘kill early’ categorydue to the failure of CMM processes to cope with difficultiesin the running of the clinical trial.

Process KnowledgeDeriving value though process knowledge is an over-archingcase of all the above elements. The automotive and electron-ics experience is that quality improvement over a multitudeof projects comes from the active involvement of suppliersrather than from ‘intervention’ and policing.

Sometimes it is possible to look at a pattern of supplydifficulties and make a single coordinated change to producea step improvement in performance. Often, it is more practi-cal to make many small, incremental improvements. Experi-ence from other industries is that this can result in realperformance benefits in the long term.

Patterns of OutsourcingAs indicated previously, other industries have had to addresscomparable issues with their own supply chains. There is agenerally recognized hierarchy of outsourcing from transac-tion-based relationships to risk sharing partnerships - Figure3.

TransactionalIn a transactional relationship, it is assumed that all suppli-ers’ offerings are comparable and so price dominates. Theoutsourcing process involves publishing invitations to ten-der, getting a number of quotes, and selecting the cheapestcredible quote.

In this relationship, costs are believed to be closely con-trolled. The reality is that significant levels of negotiation are

required, based on a contract document, to avoid creep ineither costs or requirements.

Preferred SupplierIn preferred supplier relationships, it is recognized that somesuppliers better fit the company’s requirements than others.Suppliers are selected based on some form of pre-qualifica-tion, perhaps including some elements of ‘unit pricing’ for theservices provided. Typically, a limited number of supplierswill be identified for a specified range of services.

This benefits both parties by reducing the cost of eachtransaction. The contracting company also benefits throughshortened lead-times.

In this type of relationship, the costs of services for a singleproject are less tightly controlled. However, costs are welldefined and predictable.

PartneringPartnering relationships develop where it is recognized thatthe supplier has specific competencies that complement thecontracting company’s. Clearly, before any supplier compe-tence can be described as ‘complementary,’ one must firstunderstand one’s own (required) competencies.

The partner supplier is likely to be closely involved in thespecification of work required and in the planning of projects.The cost of service is secondary although it must be related tothe value of the input. A partner supplier would typically beexpected to share in the risks of the development in some way.

AlliancesThis is appropriate where core competencies are mutuallyunderstood within a meaningful strategic framework.

Alliances tend to occur where the ‘supplier’ has key skills

Figure 3. Development of customer – supplier relationships.

Page 52: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Outsourcing Clinical Supplies

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 5©Copyright ISPE 2004

that are required by the OEM. The supplier will be solelyresponsible for certain deliverables within the project.

It is not unusual for the companies involved in an Alliancerelationship to be nominal competitors.

In Alliance relationships, the costs are a joint responsibil-ity and liability. Any profits from the venture are splitbetween the OEM and supplier according to pre-agreedarrangements.

IntegrationThis is the extreme of close customer-supplier relationships.The closeness of the relationship, and the mutual dependenceof each party, means that it is appropriate that the supplierbecomes part of the same organization as the customer. Logicdetermines mutual benefit from combining core competen-cies in a single business entity for compelling strategicreasons.

This development hierarchy is shown as a series of steps.In reality, it is more of a continuum and there can besignificant friction between the customer and supplier whenthere is a mismatch of perceptions about where the relation-ship stands.

The most important consideration is to recognize that asthe relationship gets closer, and the value invested in therelationship gets greater, so the core competencies that are

required by both parties changes.At the transactional level, the core focus for both organiza-

tions is to manage the contract. Procurement and salesdepartments are the main points of contact. In a partneringrelationship, the focus has moved away from the contract toconsider what performance improvements can be achieved byoperational changes. Line management functions becomethe main point of contact.

Planning CMM Strategic ValueHow then can companies make appropriate strategic andtactical decisions about the configuration of their CMMprocesses? The goal is to have a clearly defined frameworkthat eliminates the arbitrary nature of outsourcing decisions.The objective is to ensure that the core competencies requiredinternally are fully developed, and that qualified outsourcecapabilities are available when required.

Among the key dimensions that must be considered by thisframework are:

• The market opportunity represented by an NCE. Whatannual sales value is projected for the end drug?

• The fit of the NCE to core therapeutic areas for thecompany. What level of risk does the development repre-sent?

• the Phase of Clinical Trial being considered

Figure 4. Influence map strategic choices in Clinical Materials Management.

Page 53: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Outsourcing Clinical Supplies

6 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

These first three elements set the requirements necessary ofCMM. The business also must consider the competencies andcapabilities that are to be deployed to meet these require-ments:

• Core CMM competencies. Where does the company addmost value – managing the programs, designing studies,coordinating supplies, policing quality, producing phar-maceuticals etc?

• Available capabilities. Essentially, a combination of sup-plier management and internal performance manage-ment. In both cases, appropriate capabilities for differentClinical Trial situations must be available to the company.

The choices open can be illustrated on an influence map inFigure 4. The following statements describe how the influ-ence-map (or framework) might be deployed:

‘NCEs within the core therapeutic areas will have all activeclinical materials produced and managed in-house for allPhase I, II, and III clinical work.

CMM operations for NCE opportunities in non-core thera-peutic areas that become available to the company fordevelopment/exploitation will be outsourced from PhaseIII to our partner CRO supplier.

Our internal CMM competencies are:

• supplier quality assurance• program management• study design• bulk product production

The following activities will always be outsourced to ourPartner suppliers:

• placebo and comparator production and material man-agement

• end use packaging• logistics

The requirement for capacity in CMM will be assessedannually and any external capacity required will be placedwith partner CROs.’

Establishing such a strategic framework places some practi-cal demands on CMM organizations. They must:

• understand their core competences as determined strate-gically

• understand their effective operating capacities• operate a Capacity Planning regime that is flexible enough

to accommodate a variety of scenarios• deploy appropriate Planning and Scheduling tools to man-

age processes tactically for optimum performance

SummaryThis article describes the complex environment within whichCMM processes operate. Outsourcing is a valid mechanismfor reducing that complexity. However, any business intend-ing to outsource such processes must understand wherevalue in its CMM is derived. If outsourcing is used solely todrive down the cost of individual clinical trials, or to ‘plug’short-term arbitrary capacity holes, competitive performancewill, over time, be eroded.

Important areas of value that are embedded in the supplychain include:

• management of product and process knowledge• increase responsiveness of the organization to clinical trials• ‘portfolio’ management of new entity opportunities• fundamental competitive strategy of the business

Organizations cannot now leave the configuration of CMM asan arbitrary decision taken on a project-by-project basis.There must be clear alignment to business strategies, and afocus on developing competencies and capabilities in theresultant ‘execution’ processes.

References1. Clark, H., “Clinical Materials as Competitive Advantage,”

Pharmaceutical Engineering, September/October 2003,Vol. 23, No. 5, pp 44-52.

2. Mitchell, P., “Sending for Reinforcements,”PharmaceuticalMarketing, March 2002, http://www.pmlive.com.

3. Spehar, V., “Clinical Supplies Management,” ContractPharma, April/May 2002, http://www.contractpharma.com.

4. Miller, J., “Outsourcing Outlook,” Pharmaceutical Tech-nology, February 2003, pp 92.

5. Geimer, H. and Schulze, F., “Making Clinical Trial SupplyOperations Pay Off,” Life Science Today, February 2002.

6. Outlook 2003, Tufts Centre for the Study of Drug Develop-ment, January 2003.

About the AuthorColin Andrews completed an undergradu-ate degree at The University of Edinburgh. Hefollowed this with a period of work in theautomotive industry before going to the Uni-versity of Strathclyde in Glasgow where hecompleted a Masters degree in Business Ad-ministration. As well as line managementexperience in the automotive industry,

Andrews has 12 years of management consulting experience.He has been a consultant with SI Associates Limited for morethan six years. He has extensive experience in the pharmaceu-tical industry, both in the area of clinical trials, and in manu-facturing in Europe and North America. Andrews has under-taken a number of process improvement assignments forpharmaceutical organizations. He can be contacted by tele-phone: +44-1413325825 or by e-mail: [email protected].

SI Associates Ltd., 100 West Regent St., Glasgow G2 2QD,United Kingdom.

Page 54: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Automated Management System

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 1©Copyright ISPE 2004

Estimating ROI for Automated ClinicalMaterials Management Systemsby Lee Anderson and Devar Burbage

This articledescribes amethodologyand supportingtools focused onquantifying thecost-savingsbenefits of newsystemimplementation.

Introduction

This article describes a methodology andsupporting tools focused on quantify-ing the cost-savings benefits of newsystem implementation. The system of

interest is an automated clinical trial suppliesmanufacturing management system. The meth-odology is designed to analyze a company’scurrent operations for production of clinicalsupplies, identify those areas where an auto-mated system will have an impact, and quan-tify the expected payback in each area fromimplementing the new system. Project leader-ship will use the quantitative output of thisanalysis to help justify the project with ex-pected tangible results. Those tangible resultswill be expressed both in dollar value of theplanned identifiable improvements and as pro-

cess improvements to be implemented accord-ing to the project plan. Finally at project conclu-sion, project management will be able to com-pare actual vs. planned process improvementsas well as actual vs. planned dollar returns.

Since the amount of paperwork to do this indetail is voluminous, presented below are thedetails for one functional area only and anoverview of the results for all areas. For thoseinterested readers, the complete set of spread-sheets is available from the authors.

The results of the analysis are very positivein the sense that a large organization in aregulated environment can benefit greatly fromimplementation of an automated materialsmanagement system. This is an environmentwith many relatively small batches that needto be checked just as thoroughly as large batches.

Figure 1. Process flowdiagram forpharmaceutical researchand development.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 55: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

2 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004

Automated Management System

©Copyright ISPE 2004

Questions for Receiving Dept. #_________Areas of Savings Avg. Cost per Batch

Record1. Central handling, duplication, storage, archival $ 25

for this Dept.

Avg. Hrs/2. Time for QA to review completed Receiving 2.0

Record for Release

Avg. hrs tocorrect RR now

3. Direct Processing Time to correct Simple errors 2.04. Direct Processing Time to correct Serious errors 8.0

Other Areas of Savings FactorAvg. Ann. Value

5. $ -6. 0

Table A. Per unit questionnaire for receiving.

Although there may be small savings per batch, the largenumber of batches is the multiplier to ensure substantialsavings for this part of the project. As can be seen from theexample, we estimate cumulative net cash flow over the five-year period of this hypothetical project with an initial capitalinvestment of $1.4 million to be $4.8 million. The estimatesare very sensitive to assumptions about percent of worksaved. Using figures reported from some new systems imple-mentation projects; the estimated cumulative net cash flowcould be as high as $24.6 million.

Methodology OverviewThere are three main parts to the methodology:

1. analyze each operational area to determine the activitiesto be included in the new system

2. develop and apply a spreadsheet-based questionnaire tomeasure individual improvements in each area

3. aggregate the results in a spreadsheet to capture the valueof improvements and a final expression of the net benefitsof the project

The estimate of the net payback of the project will include thepresent value of benefits vs. expected costs over the appropri-ate time horizon and production load assumptions made bythe project team.

We take a broad approach to clinical supply materialsproduction and include drug synthesis as one of the opera-tions. Then we consider the supply chain from drug synthesisto clinical shipping as the scope of our new system. For ourpurposes, the Drug Development Supply Chain consists ofthe operational areas: Chemical Process Development, Clini-cal Manufacturing, Packaging and Labeling, and finally,Clinical Supplies Shipping.

AssumptionsEstimates of an absolute amount of cost savings are deter-mined by:

1. calculating current costs based on the volume of workunits and effort or cost per unit, then

2. calculating expected future costs as those current coststimes a work reduction factor which is an estimate

The value of the estimated reductions such as review time,archiving time, etc. that we include here is based on ourexperience with successful projects. Note: the basic value-added steps of physically receiving materials or manufactur-ing product cannot be directly impacted by an automatedrecordkeep-ing system.

ROI figures are likewise dependent on many estimatesabout future values of interest rates, inflation, and companyspecific figures such as real growth rates, and cost of capital.These will vary for each individual project and company.

Analyze Operational AreasThe implementor of a new system needs to do a detailedfunctional analysis of the activities in each operational area.The result of this is the guide to expected improvements inthe area of interest. For purposes of analysis, we begin bydividing the activities of each area into discrete functionalcategories. Depending on organizational structures, thesemay need to be added to or revised to achieve the right levelof detail. Assume we decide to deal with the PharmaceuticalR&D area as a unit. A sample process flow diagram of thatarea is shown in Figure 1. In brief, the major areas offunctionality are:

• receiving• weigh/dispense• clinical manufacturing• packaging

Develop Departmental QuestionnairePer Unit Cost Savings Estimates - ReceivingWe focus on a detailed analysis of receiving. For each activity,we prepare a questionnaire to perform a detailed analysis ofthe per-unit work to be done. An example of such a question-naire is shown in Table A, Questionnaire for Receiving, withsample values. The areas in which a new automated systemcan have an impact are those areas where excess overhead isassociated with processing records that would be electronicallystored, or areas where errors persist in the current system. Asexamples, based on unit of work, consider the time or cost to:

• Archive, Process, and Duplicate Records for a Receipt -this activity is an artifact that arises because receivingrecords in a manual system are not self-duplicating, andare not archivable in an easy manner, etc. The work is notvalued in itself, but is essential to orderly and compliantrecordkeeping for a regulated activity.

• Time for QA to Review Completed Receiving Record forRelease - the QA group has the responsibility to review allrelevant records before releasing or approving a batch ofmaterial. This time will be reduced if the receiving recordsare available electronically in report form, and can be

Page 56: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Automated Management System

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 3©Copyright ISPE 2004

reviewed for exceptions in a straightforward manner.• Direct Processing Time to Correct Simple Errors - errors in

receiving can occur for many reasons, mis-registered in-formation, wrong material classification, missing docu-ments, etc. Some errors can be classified as “simple” inthat they can be easily repaired with some minor search-ing or review.

• Direct Processing Time to Correct Serious Errors - receiv-ing errors that require detailed analysis to fix occur lessoften, but have an increased individual cost to fix.

Space is left in Table A for other areas where savings may befound – characterized as problem areas in current operations.An example is where personnel are assigned to take samplesfor analytical testing after receipt. If the sampler can’t find thematerial and no receiving personnel are available to locate it,considerable time can be lost searching for the material.

Annualizing Factors for Activities - ReceivingCorresponding to the per unit costs described above, in orderto state the values in an annualized manner, there must beestimates of the work load of the receiving department for theentire year. Consider a more or less common set of factorsthat needs to be applied to each area in Table B. The items tobe measured are:

• Average Value of Raw Materials in the Receiving Depart-ment - needed to arrive at inventory carrying cost, theaverage value is an accounting number usually availablefrom plant cost accounting.

• Number of Receiving Records in the Department pre-pared/Handled per Year - this is based on the number oflots received and the number of documents per lot, weassume 600 lots per year.

• Number of Receiving Record Errors per Year Detected -Simple

• Number of Receiving Record Errors per Year Detected -Serious

• Carrying Cost Percent for Inventory - the other part of theequation to determine inventory carrying cost, an imputedinterest rate, times the average value of Raw Materialsabove.

• Cost per Receiving Record for Central Records Handling/Archival - this is an estimate of overhead costs to properlyprocess records in the existing system.

Lastly, to determine costs associated with time spent correct-ing record and processing paperwork, estimate average loadedpersonnel costs:

• Supervisor Correcting Receiving Records• Analyst Reviewing/Correcting Receiving Records for Re-

lease• Analyst Transcribing Data from Manual Receiving Records

Determine DepartmentalExpense Reductions

Record Handling, Storage, Archiving - Receiving

Moving from paper records to electronic records, we make theassumption that we can reduce expenses for record handling,storage, and archiving by 75%. This is mainly attributable tothe fact that electronic records are by their nature easier tomanipulate, store, and archive. Given the inputs above, wecalculate the expected dollar savings for Record Handling tobe 600 records × cost per record of $25 × reduction factor of .75which turns out to be $11,250 per year.

Reduced Time to Assure Compliance(QA Release) - ReceivingWe make a similar assumption that we receive 600 lots andwe can reduce QA time to review a receiving record by 75%.The basis for this is that the receiving records and their logscan be examined on screen and they are presented in auniform format, accompanied by any associated paper records.Then, if we assume the average loaded hourly wage is $37.00per hour, the reduction in QA review time in the new environ-ment would be 2 hours per record × 600 records × .75 × $37.00= $33,300.

Replacing a system where much of the data is collected andstored on paper affords many opportunities for savings. A keycapability of an electronic system is that many of the commondata values can be predefined and when needed, the valuescan be chosen from a list. Particularly, in receiving, wherephysical lots of material from diverse suppliers and manufac-turers are processed, any automation of data entry helpsreduce errors. We estimate that 90% of the data entry errorscan be eliminated by a system where many common valuesare predefined and only selection from a list can be done atprocessing (e.g., receiving) time. Referring to Tables A and B,we see of the 600 receiving records, 100 are estimated tocontain simple errors to be fixed and 25 are expected to haveserious errors that need to be fixed. Each fix requires time.We estimate the cost of a supervisor correcting the problemsto be $40.00 per hour. Then the cost savings associated withbetter data entry and reduced errors can be calculated as the

Common Dept. Factors Needed for the Analysis:1 Name of Department2 Type of Dept (Receiving, Weigh/dispense, Mfg, R

Pack, and Other depts.3 Average Value of RM in Receiving Dept. $ 500,0004 # Receiving Records in Dept. prepared/handled $ 600

per year (lots)5 # Receiving Record Errors/year detected - 100

Simple (# errors)6 # Receiving Record Errors/year detected - 25

Serious (# errors)7 Carrying Cost percent for inventory (percent) 208 Cost per receiving record for Central Records $ 25

handling/archival9 Average Loaded Hourly Personnel Cost:10 Supervisor correcting Receiving Records $ 4011 Analyst reviewing/correcting Receiving $ 37

Records for Release12 Analyst transcribing data from manual $ 36

Receiving Records

Table B. Annualizing questionnaire for receiving.

Page 57: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

4 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004

Automated Management System

©Copyright ISPE 2004

cost savings associated with the simple errors ($7200 = 2hours per fix × .90 savings × 100 records × $40 per hour) plusthe cost savings associated with complex errors ($7200 calcu-lated in a similar manner).

Reduced Inventory/Material Savings(Variable Overhead) - ReceivingThe other main cost area in receiving is the value of inven-tory. We postulate that we can reduce the amount of inven-tory carried by taking advantage of the improvedrecordkeeping capabilities of the new system to:

• improve material throughput• reduce loss of material due to obsolescence• avoid retesting and time needed to expedite items

Then we expect we can reduce the average inventory over ayear’s time by 40%. In this case, where we expect the averageinventory to be $500,000, that means a one-time savings of$200,000, followed by an annual cost savings of $40,000 (20%of that $200,000).

Departmental Financial Benefits -Aggregate the Above BenefitsIf we add the savings from each source above, we get totalannual savings for the receiving department calculated to be$98,950 and the one-time savings from a reduction in inven-tory is $200,000.

Overview of Aggregate Expense ReductionsGiven assumptions similar in nature to the ones made forReceiving, we have calculated estimated savings for the othermodel areas in our business. Table C shows a summary ofthose savings.

• Weigh/Dispense - Of particular note are the labor savingsof the Weigh/Dispense Area, estimated to be $291,581.The Weigh/Dispense area is assumed to carry a work loadof 1500 dispense orders per year. The savings are obtainedmainly by assuming 20% reductions in manual reportpreparation and duplicate data entry, and a 25% reductionof direct labor spent only on verification. A more aggres-sive assumption of 75% reduction in manual report prepa-

ration and 100% elimination of time spent on verificationwould yield savings of $1.2 million.

• Manufacturing - There is a large savings in materials inthe Manufacturing Department, which we assume to bemaking 1500 batches of product a year. The savings aremainly attributable to reductions in throwaway due toProcess Error and Incorrect or lost batch record paper-work, along with savings by reducing waste due to a toohigh overage allowance on weighing.

• Packaging - The Packaging Department is estimated tobe saving $637,300 per year in labor. The workload isassumed to be 7800 packaging orders per year. A largepart of the labor savings is based on the same assumptionas in receiving that there is a 20% reduction in the time forQA review. Other savings come from 20% reduction intime spent locating material, elimination of time spentwaiting for material delivery, etc., and lastly, a 25%reduction in packaging order preparation time estimatedat 2 hours per order reduced to 90 minutes. As above, if weare more optimistic about the reductions in effort with thenew system, we can project savings of up to $2.5 million.

Multi-Year ROI - Overview of AnalysisIn Tables D and E, we present the final analysis sheets. InTable D, we present Capital Investment and Expense Reduc-tion. The capital investment figures are straightforward.Expenses and Expense Reductions are based on a 10% realannual growth rate and a 3% inflation rate. Note that dis-placed costs of maintenance on existing systems also count asa saving. Taking a conservative approach to the savings to beexpected over the 5-year period, and adding across, the totalcapital investment is $1.4 million and the Expense Reductionis $10.6 million.

Table E takes account of the extra expenses incurred dueto the new system – software support, platform maintenance,etc. Net income is based on the project income and expensefigures (pretax income) less taxes at an assumed 40% taxrate. Finally, we see cash flows from project operations goesfrom -$127,500 in year 0, when there is no savings to be had,to $1.5 million in year 5.

Annual SavingsFrom Individual Workbooks One Time Savings Labor Mat’l Var OH Other Dir Cost

1a. Receiving Department 1* $ 200,000 $ 11,925 $ 10,000 $ 2,8132a. Weigh/Dispense Department 1 $ - $ 291,581 $ 16,250 $ - $ 10,2813a. Manufacturing Department 1 $ 299,962 $ 187,780 $ 387,500 $ 23,058 $ 34,8444a. Packaging Department 1 $ 591,771 $ 637,300 $ 15,750 $ 52,023 $ 42,188

Totals Labor Mat’l Var OH Other Dir CostAnnual Savings $ 1,128,586 $ 419,500 $ 85,080 $ 90,125One Time Savings $ 1,091,733

* = reduced expected labor savings Annual Savings Grand Total $ 1,723,291

Table C. Aggregate savings across all departments.

Page 58: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Automated Management System

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 5©Copyright ISPE 2004

SummaryTo summarize the steps to follow, model a series of manufac-turing activities to develop a solid estimate of the expectedROI for the implementation of an automated clinical suppliesmanagement system. First, determine the scope of the project– in this case, the drug development supply chain. Then,break down the supply chain to operational units that can beexamined separately or are organizationally separate. Per-form a detailed functional analysis of the activities andgroups engaged in each functional area. On the basis of thedetailed analysis, develop focused spreadsheet question-naires, gathering information on per unit costs and effort;and on total number of units and effort. This is the basis forthe estimate of current costs.

With the spreadsheets and assumptions about reductionof effort using the new system, and the work redefinition thataccompanies that, one can estimate the new levels of costsand effort. The difference between the new and old figures is

the amount of gross savings per year. Then one can project thegross savings per year along with estimated project costs,estimated tax rates, inflation rates, etc. to arrive at the finaldetermination of ROI.

The actual ROI figures are one part of the total informa-tion requirements needed to decide if a project should bepursued. In a regulated environment, assessments must bemade as to the validation effort required to put the system inplace and the ability to maintain demonstrable control overthe functioning system once it is in place. A proper analysisfor purpose of determining ROI can yield information that isapplicable for many other purposes.

The assumptions about reductions in materials, reducedeffort for QA review, and reduced errors while performingactivities are key to what to expect from installation of a newsystem. Expectations for returns on a project, assuming nomajor changes in workload, are based on these assumptions.Since they are so critical to the decision making process, a

Table D. Return on investment - 5 year view - capital investment and expense reduction.

Year 0 Year 1 Year 2 Year 3 Year 4 Year 5 TOTALI. CAPITAL INVESTMENTSystem Costs

SOFTWARE -500,000 0 0 0 0 0 -500,000SERVICES -800,000 0 0 0 0 0 -800,000

PLATFORM: HARDWARE/SOFTWARE -150,000 0 0 0 0 0 -150,000REDUCED WORKING CAPITAL 0 0 0 0 0 0 0TOTAL -1,450,000 0 0 0 0 0 -1,450,000

2. EXPENSE REDUCTIONAssumed CAGR 10%From Dept. Spreadsheets

Labor 0 1,128,586 1,241,445 1,365,589 1,502,148 1,652,363 6,890,130Materials 0 419,500 461,450 507,595 558,355 614,190 2,561,089Variable Overhead 0 85,080 93,588 102,947 113,241 124,566 519,422Other Direct Costs 0 90,125 99,138 109,051 119,956 131,952 550,222

Displaced Costs of Current System Maint 0 10,000 11,000 12,100 13,310 14,641 61,051TOTAL 0 1,733,291 1,906,620 2,097,282 2,307,010 2,537,711 10,581,915

Table E. Return on Investment - 5 year view - expenses, depreciation, taxes, and net income.

Year 0 Year 1 Year 2 Year 3 Year 4 Year 5 TOTAL3.INCREASED EXPENSES

Assumed Rate of Inflation 3POMS SW SUPPORT -85,000 -87,550 -90,177 -92,882 -95,668 -98,538 -549,815PLATFORM MAINTENANCE -22,500 -23,175 -23,870 -24,586 -25,324 -26,084 -145,539SYSTEM SUPPORT PERSONNEL -80,000 -25,000 -25,750 -26,523 -27,318 -28,138 -212,728INCREMENTAL CLIENT SERVICES -25,000 -25,000 -25,750 -26,523 -27,318 -28,138 -157,728TOTAL -212,500 -160,725 -165,547 -170,513 -175,629 -180,897 1,065,811DEPRECIATION EXPENSE 0 -290,000 -464,000 -278,400 -167,040 167,040 1,366,480PRETAX INCOME -212,500 1,282,566 1,277,073 1,648,369 1,964,342 2,189,774 8,149,624COMBINED FED/STATE TAX RATE 40%INCOME TAXES 85,000 -513,026 -510,829 -659,348 -785,737 -875,910 -3,259,850NET INCOME 127,500 769,540 766,244 989,021 1,178,605 1,313,864 4,889,774DEPRECIATION ADD-BACK 0 290,000 464,000 278,400 167,040 167,040 1,366,480CASH FLOWS FROM OPERATIONS -127,500 1,059,540 1,230,244 1,267,421 1,345,645 1,480,904 6,256,254CAPITAL INVESTMENT -1,450,000 0 0 0 0 0 -1,450,000NET CASH FLOWS -1,577,500 1,059,540 1,230,244 1,267,421 1,345,645 1,480,904 4,806,254

Page 59: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

6 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004

Automated Management System

©Copyright ISPE 2004

useful exercise is to identify each assumption, and carry theassumptions as entries in an “independent variables” spread-sheet where changes can be made easily, then by increasingor decreasing one or all of the assumptions, a sensitivityanalysis can be made. If, when a key assumption is changedby 10%, the corresponding results change by 20% or somelarge amount, the assumption needs to be examined in detailto get the necessary level of assurance that it is correct. Whenoffering an in-depth analysis of expected ROI, a range ofexpected values - high savings, expected savings, or lowsavings should be made.

Further ExtensionsThe discussion above focused on the analysis associated withcost savings, almost entirely from labor or materials. Themodel can be extended to consider changing workloads,scaling as appropriate. Other advantages from implementa-tion of a new system are harder to quantify. If the QA groupascertains that the new system provides more visible compli-ance or reduced effort to determine that batches can bereleased or not, a real value exists, but may be hard toquantify in dollar terms. If the new system is part of anoverall Business Process Re-engineering effort the total ex-pected returns might be hard to separate from the expectedreturns from the system itself. The analysis that precedesimplementation of a new system should consider all expectedadvantages and disadvantages, quantify wherever possible,and report any other items that should be considered.

About the AuthorsLee Anderson has more than 30 years ofexperience in commercial and plant floorsystems. He recently joined the HoneywellPOMS organization as Product Manager, re-sponsible for MES and CMS products. He hasheld project lead and project managementpositions at DuPont and DuPont Pharma-ceuticals. He has prior experience in Auto-

mated Materials Handling, Planning and Scheduling, Scien-tific Modeling, and in Banking Systems. He has degrees ineconomics, computer science, environmental science, andaccounting. He can be contacted by tel: 1-703/793-4460 or bye-mail: [email protected].

Devar Burbage has more than 30 years ofexperience in plant floor systems. He has heldpositions in marketing, sales, systems engi-neering, systems development, and consult-ing. Companies include Honeywell, IBM,Manugistics, and GRC International. His ar-eas of application expertise include opera-tions management, MES, ERP, advanced plan-

ning and scheduling, plant maintenance, environmental healthand safety, and process automation. He recently retired asdirector of alliances for Honeywell-POMS; a Honeywell busi-ness unit bringing process management solutions to the FDAregulated industries.

Honeywell POMS, 13655 Dulles Technology Drive, Suite200, Herndon, VA 20171.

Page 60: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Annex 13

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 1©Copyright ISPE 2004

Annex 13: An Updateby David Barnes

This articleprovides acurrentoverview of theEuropeanCommission’s(EC) GoodManufacturingPractices(GMPs) “Annex13.” Annex 13is used for themanufacture ofinvestigationalmedicinalproducts. Thisoverview willinclude a briefhistory of theAnnex, therelationship theAnnex has withthe ClinicalTrials Directive,and acomparison ofthe 2003version with thecurrent (2002)version.

Background History

In the past, the European Union (EU) mem-ber states have had no legal requirementsregarding the manufacture or packaging ofmedicinal supplies for clinical studies us-

ing GMP guidelines. The majority of compa-nies have strived to achieve GMP guidelines inthe clinical manufacturing and packaging ar-eas. Many of the GMP guidelines used werefrom the commercial sectors of the company orwere taken from the United States Code ofFederal Regulations (CFR).

By the late 1980s, the EC drafted two direc-tives using GMP guidelines for the manufac-ture, packaging, and storage of medicinal prod-ucts. The first directive addressed medicinesfor human use (91/356/EEC). The second direc-tive addressed veterinary products (91/412/EEC). While these directives covered commer-cial products, they also contained a clause thatallowed countries in the EU member states torequire that “clinical study materials” also beprepared according to GMP. This can be consid-ered the origin of Annex 13.

The adherence to the GMP guidelines forclinical trials materials will become manda-tory with the implementation of the ClinicalTrials Directive (2001/20/EC). To facilitate or-ganizations with guidance in preparing sup-plies for clinical studies, in accordance with theGMPs, an extra Annex to the document (en-titled “The Rules Governing Medicinal Prod-ucts in the European Union - Volume 4” orcommonly called the “Orange Guide”) was writ-ten. Annex 13, specifically addresses the differ-ences in practices when making investigationalproducts, as compared with commercial manu-facture and packaging. It provides a guideagainst which facilities preparing clinical ma-terials can be audited. GMPs not covered inAnnex 13 can be found in other sections of the“Orange Guide.”

Annex 13 was revised and published on 25July 2003. It is available for review at the

following Web site: http://pharmacos.eudra.org.The content of this article references the 2002and 2003 versions of Annex 13. Each versioncontains 13 sections. However, the format forthe two publications is different. The format-ting in this document is organized to follow atypical workflow pattern.

IntroductionThe “Introduction” section of the 2002 versionprovides the historical background to Annex 13(similar to the “Background History” sectionprovided above). In the 2003 version, this sec-tion is referred to as “Principle.” The “Prin-ciple” section no longer provides historical de-tails; however, it clarifies the status of productsother than the test product used in a clinicaltrial, e.g., a diagnostic agent. These detailsshould be an appropriate quality for the pur-poses of the trial. The reader is advised toobtain the advice of a Qualified Person (QP) toassist in the process. A new addition to the 2003version is a useful glossary of terms.

Quality ManagementBecause investigational products are rarelyvalidated to the standards required for mar-keted products (except as required for steriliza-tion cycles) and these products have specifica-tions and manufacturing processes that fre-quently change, Annex 13 requires both aneffective and well-documented Quality Assur-ance system. The primary difference in the2002 edition versus the 2003 edition is thenotation that the packaging and labeling pro-cesses for clinical trial supplies are differentwhen compared to those for commercial prod-ucts. This invokes the EC Guideline on GoodClinical Practices (GCPs) requirement for self-inspection and independent audit. This nota-tion is absent from the 2003 edition.

PersonnelThe 2002 edition of Annex 13 does not specify

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 61: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Annex 13

2 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

the need for the involvement of QPs in the release of investi-gational products. What it does state is that the people whoare responsible for Quality Control/Release of investiga-tional products should be trained in the appropriate GMPsand independent of the persons who manufactured the prod-uct. However, this omission was addressed in the 2003edition. It clearly states in a comprehensive list the dutiesand responsibilities for the QP(s). The QP is responsible forensuring that there are systems in place that meet Annex 13requirements. Additionally, there must be a thorough under-standing of clinical trial processes and drug development.The QP is responsible and legally obligated for the quality ofthe released products.

Premises and EquipmentBoth versions of Annex 13 recognize the difficulties associ-ated with validating aseptic processes for small batch sizesand suggest that environmental monitoring be enhanced inthese cases to provide the necessary assurance. The 2003version of Annex 13 suggests that the media fill runs of abatch size larger than the clinical trial materials batch becarried out “to provide greater confidence.” However, noguidance is provided in determining the acceptable resultsfor these media fill runs.

When working with sensitizing products such as ‘penicil-lin’ or ‘toxic/highly potent’ drugs, the current edition of Annex13 removes the need for dedicated facilities. It does empha-size the need to thoroughly decontaminate the equipmentand facilities used. There is no mention of this subject in the2003 revision.

DocumentationAnnex 13 reflects an understanding that the technical speci-fications for investigational products will be modified duringthe course of the product’s development. Annex 13 requiresthat all changes made be recorded. Specific references toprevious versions of the product specifications along with therationale for the change must be recorded in the documenta-tion.

Clear written instructions for manufacture and supplymust be produced. Additionally, written records of the opera-tions performed must be kept. The 2002 edition requires thatthese records must be kept at least two years after the end ofthe clinical trial. Unlike commercial manufacture, there is norequirement for master formulae and processing instruc-tions. A written procedure that incorporates the modifica-tions to the product as well as the authorization of the productmust be established. The 2002 edition requires that particu-lar attention be given to product stability and bioequivalence.

The sponsor of a study should provide an official writtenorder or may transmit an electronic order to the manufac-turer. It should be precise, and in the 2002 version, it isrequired to refer to the order as the ‘Product Specification’file. The 2003 version acknowledges that the order mayadditionally need to be referenced to the ‘Clinical TrialProtocol’ if necessary. Both versions of Annex 13 require thatinvestigational products have a ‘Product Specification’ file

similar to that for commercial products. This should includeformulation, processing, packaging, testing, storage, andshipping information. As suggested above, this documentwill need to be updated as changes occur and these changesneed to reference previous versions. While the 2002 versionprovides little guidance in this matter, the 2003 revisionprovides a list of required documents and information. It alsoprovides that where a number of sites are involved in theoperation, each site can maintain a file pertaining to thatsite’s operations.

Annex 13 recognizes that the packaging and labeling ofinvestigational products is generally more complex than forcommercial products especially when “blinded” labels areused. Therefore, it requires that processes such as labelreconciliation and line clearance be appropriately enhancedand performed by an independent Quality Control staff.

Annex 13 provides allowances for batches of investiga-tional products that may be packaged as a number of sub-lotsover a period of time. The number of units to be packed eachtime must be specified beforehand allowing for quality con-trol and reserve samples. Detailed reconciliation of the pack-aging and labeling processes must take place.

Labeling InstructionsAnnex 13 provides detailed labeling requirements for boththe outer and primary packaging of investigational products.It allows for the use of symbols and pictograms on the outerpackaging. Copies of the labels should be retained in thepackaging records.

There has been much discussion about the need to providea ‘period of use.’ The 2002 version of Annex 13 addresses theneed for ‘in-use extensions’ to ‘expiry dates of investigationalproducts’ by requiring that expiry extensions should be onadditional labels. These labels may be affixed over the origi-nal expiry labeling. The extra labels must include thematerial’s batch number and when affixed to the containermust not obscure the original label’s batch number. Althoughnot stated, this means that if multiple expiry updates aremade, the labels should be designed to obscure previousexpiry dates and leave all examples of the batch numberclearly visible.

The 2002 version states that the clinical trial monitor orsite pharmacist must apply these labels according to writtenSOPs. A second person must verify the application of thelabels. This must be documented in the trial documentationand batch record. The 2003 version introduces the need to dothis at the manufacturing site and provides some flexibilityto be performed at other sites as required.

The 2003 revision has not addressed many of industry’sconcerns over labeling requirements. The original list ofrequirements remains, but the revision offers an opportunityfor the contents of the labels to be reduced by stating (prior tothe list of requirements) “The following information shouldbe included on labels, unless the absence can be justified, e.g.,use of a centralized electronic randomization system.” Fi-nally, the 2003 revision provides a useful summary of label-ing details required for various sized containers.

Page 62: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Annex 13

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 3©Copyright ISPE 2004

ProductionAnnex 13 requires that the starting material qualities needto be defined and periodically reevaluated. The specificationsfor the active materials should be as comprehensive aspossible. This information on material quality should bemaintained to allow for variations in production.

Annex 13 recognizes that validating manufacturing pro-cesses to the standards required for marketed products is notappropriate for clinical trials supplies, and, because of this,quality control testing takes on more importance. Annex 13requires that provisional parameters and in-process controlswill be put in place, possibly based on experience withanalogues. These processing procedures, parameters, andcontrols should be adapted in light of experience. There is aspecific mention of reconciliation in Annex 13 with the re-quirement that it is carried out and that any abnormalresults be investigated.

In recognizing the potential of a product’s toxicity, Annex13 specifically states that the cleaning procedures employedto decontaminate facilities and equipment should be strin-gent. Where biological materials are used, impurities mustbe removed with the same stringency as for marketed prod-ucts.

Comparator AgentsAnnex 13 requires that the quality and integrity of compara-tor agents should not be compromised. If significant changesare made to comparator agents, data must be available todemonstrate that these changes have no significant effect.The 2002 version states that the expiry date on the originalpack of comparator is based upon the original packaging andthat it may not be appropriate when repackaged. Therefore,the sponsor company has to provide data to support expirydate in any new packaging and specify that it cannot be laterthan the date of the original pack. If no data is available, thenew expiry must be less than 25% of the period from repack-aging to original expiry date, or six months, whichever is less.The 25% requirement is not listed in the 2003 version. Thestatement now reads as “there should be compatibility ofexpiry dating and clinical trial duration.” The 2003 revisionis less specific and more ambiguous.

A specific requirement in Annex 13 states that all aspectsof the generation and subsequent handling of randomizationcodes must be in a procedure and that there needs to be aprocedure to identify blinded products. This procedure andthe randomization code must allow complete traceability ofproducts to the original batch number of the product beforeblinding.

Quality ControlIn the absence of full process validation, end product testingis very important. The testing should assess those character-istics affecting the product’s medicinal efficacy, e.g., dose anduniformity, release of active ingredient, and estimation ofstability. Where appropriate, these tests should cover thecharacteristics of blinded products.

Batch ReleaseThe text in the 2002 version recognizes two processes: therelease of bulk product and the final packaged product. Itrequires that bulk product testing demonstrate compliancewith the specifications listed in the ‘Product Specification’file. Finished (i.e., packaged) product testing should be com-pliant with the specifications and include a review of packag-ing documentation.

The 2003 version significantly expands this section, and inparticular, provides detailed guidance on the duties of a QPin this matter. One interesting aspect of this is that a QPreleasing materials from a non-EU country (referred to as athird country) will need to determine that standards of GMPequivalent to those in Annex 13 have been applied. One wayof achieving this is participating in an audit of thatmanufacturer’s quality systems. One consequence of thisis that the QP of a EU-based Contract Research Organization(CRO) packaging materials for a customer manufactured ata third country site may audit that site to ensure that thequality systems are appropriate. So, not only will the cus-tomer audit the CRO, but quite probably, the CRO will haveto audit the customer.

Free MovementAnnex 13 clearly states that once appropriate testing andrelease has occurred within the EU, there is no justificationfor any further testing as long as the correct control proce-dures have been followed and documented.

Contract Manufacture and AnalysisThe 2002 edition of Annex 13 specifically references contractmanufacture and analysis requiring the contracting partiesput a contract in place clearly stating that the products are for‘clinical trial use only.’ There is no reference to this in the2003 revision.

ComplaintsAnnex 13 requires that all parties discuss conclusions ofinvestigations from any complaints. This helps to assess theimpact upon the clinical trial and product development. The2003 version specifically requires the involvement of the QPin this process.

Recalls and ReturnsBoth versions of Annex 13 require procedures to obtain anddocument the retrieval of materials. The investigator mustfully understand and monitor these procedures. The 2003revision adds another requirement for the sponsor to ensurethat the supplier of a comparator agent has a system foradvising the sponsor in the event that there is a need to recallany products.

Shipping, Returns, and DestructionAnnex 13 requires that a shipping order be produced in orderto ship materials. The shipment can only be made whenquality control releases the product and then is approved bythe Regulatory department. Release from both groups should

Page 63: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

Annex 13

4 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

be recorded and retained. The materials must be packaged toensure that the product is protected during transport andstorage, and that opening or tampering of the package iseasily identified. An inventory of the shipment should bemade and retained including the addressee information. Afinal requirement is the acknowledgement of a receipt in goodcondition. The 2003 revision adds a need for decoding ar-rangements to be available prior to the materials beingshipped to the investigator site.

Shipments from one trial site to another should be avoided.If such transfers must be made, a detailed procedure of theprocess is required. Annex 13 specifically states that it ispreferred that a product is returned to the sponsor for relabel-ing and possible retesting before being sent to the second site.Materials that are returned to the sponsor for whateverreason should be returned in a specified manner, be clearlyidentified, stored in a dedicated area, and accounted for ininventory.

Annex 13 clearly states that the sponsor is responsible forthe destruction of all materials. Destruction of the materialsshould be recorded and carried out only after completion ofthe trial and compilation of the final report. Where a partyother than the sponsor carries out destruction, that partymust provide documentation detailing the identity of thematerials destroyed and the quantities. The 2003 revisionrequires that the records of destruction be held by the spon-sor. This is not a typical GMP practice because the sitecarrying out the activity holds the original documents.

ConclusionAnnex 13 attempts to deal with clinical supply specific issuesthat the “commercial” GMPs either do not address or do so in

an inappropriate manner. Obviously, there are statements inAnnex 13 that many will disagree with, but it should beremembered that these are guidelines, not rules, and that asexperience of auditing clinical supply facilities is gained, theywill undoubtedly change.

About the AuthorDavid Barnes graduated from The School ofPharmacy, University of London with abachelor’s degree in pharmacy in 1983 with afinal year specialization in pharmaceuticalengineering. In 1988, he obtained a PhDfrom the same institution and joined Pfizerin Sandwich, England. Starting out in prod-uct and process development, Barnes devel-

oped manufacturing processes for tablet formulations andtransferred them into production facilities. He then joinedthe design and validation team for the Pharmaceutical Sci-ences Building in Sandwich, leading the clinical manufactur-ing group when the building opened. In 2001, Barnes trans-ferred to the Pfizer facility in Ann Arbor, Michigan to com-plete the commissioning and validation of the site’s newTechnical Development Facility’s GMP areas. Barnes is nowthe head of the Technical Support and Innovation Group inMichigan Pharmaceutical Sciences. He is a member of theRoyal Pharmaceutical Society of GB, the Institution of Chemi-cal Engineers (UK), and ISPE where he is a member of theClinical Materials Committee. He can be contacted by e-mail:[email protected].

Pfizer Inc., 2800 Plymouth Rd., Ann Arbor, Michigan48105.

Page 64: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 2

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 1©Copyright ISPE 2004

External Sourcing of Clinical TrialMaterialsPart 2: Impact of Electronic Automation on theSourcing Model for Clinical Supplies Preparationby Charles F. Carney

This articleidentifies anddiscusses someof the regulatoryand controlsystemconcerns for theutilization ofelectronicsystemsbetweensponsor andcontractorcompanies forthe preparation,control, anddistribution ofclinical trialmaterials, andprovides someideas for theproactivedevelopment ofthe relationshipbetweensponsor andvendor that canminimize anyadverse impactand maximizethe strengthsand value of therelationship.

Introduction

The business practices models for sourc-ing of clinical trial supplies preparationhave traditionally relied on personalinteractions to link the manual informa-

tion and control systems of the sponsor andvendor. Even when electronic systems existedin both sites, these have traditionally beenutilized only within that site, relying on manualinteractions between sponsor and vendor forthe transfer and review of information, data,and records, all in paper medium. In the future,we will see a change in paradigm toward areliance on electronic interface, preparation,storage, retrieval, and review of information,data, and records. This change is required bythe call for increasing efficiency and decreas-ing timelines for drug development. Some pointsthat will be considered in this article are:

• the quality system points in the cGMP of theUS and EU that need to be taken into ac-count, including specific points in 21 CFRPart 11 and Annex 11 respectively

• achieving harmonization of concept and prac-tice, between the sponsor and vendor, forinterpretation of the regulatory require-ments for qualification, validation, control,and interchange of information betweenseparated electronic systems

• comparison of concerns when the electronicautomation system is the same in the spon-sor and vendor sites and when the electronicautomation systems are different

• optimizing the interactions between thesponsor and vendor for the development,approval, execution, archiving, and retrievalof electronic automation records and data to

ensure accountability, responsibility, andconsistency

The Use of Electronic SystemsWhen the computer age arrived for pharmaceu-tical products, many thought that electronicautomation could surely improve efficiency,effectiveness, and throughput within an orga-nization, but could not be utilized betweenorganizations. This resulted from the fear thatdifferences in the computer concepts withineach organization could potentially hinder orinvalidate the interactions. Most felt that vali-dation of the interface would only be possible ifboth organizations were utilizing the same com-puter system, configured in the very same way.Anticipating a very low probability for the ex-istence of the exact same computer system andconfiguration in both organizations, mostthought that the costs for validation of thelinked computer systems would far outweighthe cost advantages of sourcing the work. There-fore, most have utilized manual systems fortracking and exchanging information and datawhen sourcing investigational materials froma contract manufacturer. However, in order tobest leverage a sourcing relationship, one musttake advantage of each efficiency and effective-ness factor in both organizations. Therefore,one must look for the ways to connect andutilize the electronic systems in the two organi-zations, to perform the evaluation of compli-ance, and to define the points of interactionbetween the two systems in such a way that allof these features can be accepted and utilizedfor the benefit of both the sponsor and thevendor.

Information for the preparation of clinical

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 65: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 2

2 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

materials includes: data in any of the manufacturing records,information for the receipt, holding, and distribution of prod-ucts and associated reserve samples, information for thetesting for release and stability evaluations of products,materials specifications, label specifications and text, infor-mation for any comparator products or ancillary medica-ments or diagnostic goods for the trial, and all evaluationsand reports for Out of Specification (OOS) results, deviations,and other quality investigations. One also could consider theGood Clinical Practices (GCP) information necessary for theexecution of the trial at each of the sites, for example: casereport forms, inventory and dispensing records, informedconsent forms, protocol deviation reports, field complaints,etc. However, this article will consider the Good Manufactur-ing Practices (GMPs) issues and leave the consideration ofthe GCP issues as an exercise for the reader.

The optimal characteristics for such a combination andutilization of the two electronic systems for the manufactur-ing, packaging, labeling, and distribution of clinical suppliesmust contain elements of quality control, quality assurance,and validation that are understood for each of the systemsworking alone. Thus, the two companies can focus on ensur-ing that the interface between the two systems does notinterfere with the existing capabilities and compliance ofeach system acting independently.

Quality and Management SystemsMost people recognize the important elements of cGMPcompliance. Therefore, this discussion will only emphasizesome of the points to consider for the utilization of electronicautomation systems by sponsors and vendors in a workingrelationship. The requirements for computer systems con-tinue to evolve, and therefore, this presentation is intendedto be more conceptual than definitive. The recent withdrawalby the FDA of guidance documents for application of 21 CFRPart 11 is evidence that this regulatory agency is refining themeaning of compliance for electronic systems.

All regulatory systems recognize the increased consis-tency when properly qualified and validated electronic auto-mation is utilized. However, these systems also recognize thevalue of human observation and recognition followed byintervention for something that is “not quite right.” Thereforeall compliance systems which will utilize electronic automa-tion also should ensure that no loss in product quality,through loss of human oversight, may occur by the utilizationof such systems.

An important aspect of all business and compliance sys-tems which should never be overlooked or ignored is the roleand responsibility of management personnel.1-3 Just as themanufacturing management personnel within an organiza-tion must understand and rely on their counterparts in thecomputer support, analytical, and purchasing departments,so should these personnel know and understand the workingof their manufacturing counterparts and those in the sup-porting departments in the vendor company as well. Thisrequirement is implied in the US regulations and stated moreexplicitly in the EC regulations.4 This combined manage-

ment comprehension and joint execution of oversight bringsadded strength to the sourcing relationship.

The key aspects of electronic automation can be statedhere with elaboration only as needed to support the ideas ofthis presentation. These electronic systems include, but arenot limited to, computer aided systems for inventory manage-ment, manufacturing processes, analytical release and sta-bility data management and reporting, label generationincluding randomizations and text, and excursion reportingand archiving. These systems support the quality manage-ment decisions made for each product produced. Because theelectronic aspects are so important, and because there shouldbe no confusion about these aspects, they are given specialtreatment within the regulations in the US5 and in the EU.6

The former is more detailed and has had associated with it,until recently, several separate guidance documents for in-terpretation. However, both need to be considered to have acomplete compliance program suitable for allowing actions inthe world.

The following special points need to be considered in orderto assure that the link between the companies has takenthem all into account for full compliance:

• detailed written description of systems and the linksbetween them for the two companies, security, how theyinteract, who has authorization to perform what tasks ineach system, quality system(s) oversight, and interactionsfor any changes

• assurance that data entry has been checked and that thedata are consistent between the two companies (in theirdatabases and in the archives)

• When first initiating use of electronic automation a checkshould be made, by running in parallel, that the electronicautomation system and the manual system provide thesame degrees of control and reliability of the information.

• Agreement should be reached concerning who can input,amend, or alter data and agreement on the process andcommunication of changes in authorization level withrespect to data.

• The change control system needs to take into account theapproval by the two quality assurance systems and the re-education of all concerning the changes. This includes anyexercises for re-qualification, re-validation, and anychanges in organizational structure.

• Security of stored data should be agreed and accessibility,durability, and accuracy of the stored or archived datashould be evaluated periodically. The files backup planshould be agreed between the two entities. Additionalpoints specified in the European regulations include:7

- Detailed procedures should be in place and availablefor review.

- Only authorized personnel should manipulate the elec-tronic records and associated data and all transactionsneed to be clearly tracked by an electronic audit trail.

- Back-up copies of records, on durable medium (mag-netic tape, microfilm, paper or other) should be pro-duced and archived.

Page 66: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 2

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 3©Copyright ISPE 2004

- All data and controlled information should be readilyavailable throughout the retention period.

A disaster recovery for failures and break-downs and alter-nate systems usage in times when the primary system be-comes not usable must be agreed prior to initiation of theinteraction.

A written agreement (contract) must always be in forceduring the relationship to avoid misunderstandings whichcould result in products of less than acceptable quality and alldetails need to be included specifically. Some of the criticalpoints listed in the European regulations are:8

• The contract must state clearly the responsibilities andduties of each party.

• The contract must identify the pathway followed by theQualified Person (QP) for releasing a batch.

• Any special technical arrangements should be specified.• Information exchange should be complete and sufficient to

ensure that the vendor is able to perform the requestedwork successfully and compliantly.

• All materials must comply with their specifications and bereleased for use by the QP.

• Use of subcontractors for work must be pre-approved bythe sponsor.

• Materials purchase, product testing including samplingand in-process control testing, and release responsibilitiesmust be specified clearly and accepted by both parties.

• Audits by sponsor and/or competent authorities must beclearly understood and accepted by the vendor.

• Annex 13 of the EC Guide also specifies that the contractmust clearly state when product will be utilized in clinicaltrials both for manufacturing and for testing operations.9

Agreement for the QP must be in place and assignments ofauthority and responsibilities must be clearly written accord-ing to the requirements of the European regulations.10 Someof the critical points are:

• How the QP of the sponsor takes over the decisions of thevendor and which decisions the QP of the sponsor makesbased on work performed by the sponsor must be clearlydefined and written.

• The QP, whether in the sponsor or the vendor firm, mustbe sufficiently knowledgeable of the specific nature of theinvestigational drug and its control to be able to performthe full functions required of the QP.

• An obligatory notification procedure must be specified forreporting of any aberrant data by the vendor to thesponsor.

• Any computer system used must meet the requirements ofAnnex 11.

• The identification and withdrawal of any product lotfound, after release, to be hazardous because of a qualitydefect, must be assured and must occur without delay.

The quality system points in the US cGMPs, whether codi-

fied11 or exemplified by inference12 as a “customary practice,”and codified in the GMP of the European Union,13 andexemplified by inference in the rest of that document and itsannexes need to be taken into account. These include specificassignments of personnel to the control duties specified forthe quality control unit in the US and the production andanalytical control personnel assignments of the EU. Theadditional points for clinical supplies stated in Annex 13 forquality assurance and quality control must be considered inany quality system for the preparation of clinical trial mate-rials. This is particularly true for blinding, labeling, valida-tion, distribution, returns, and destruction procedures andcontrols.

Harmonization of Concepts -Sponsor/Vendor

Very deliberate efforts and negotiations are required toachieve harmonization of concept and practice between thesponsor and vendor for interpretation of the regulatory re-quirements. This applies to manual systems as well as toelectronic automation systems. Qualification, validation, con-trol, and interchange of information occur between personnelas well as between separated electronic systems.

First and foremost both the sponsor and the vendor needto have a concept for gaining the understanding of thesystems in the other site. This is relatively easier for thesponsor because the sponsor will usually visit a vendor sitefor technical and quality evaluations as part of the duediligence activities prior to proposing a contract relationship.The vendor seldom has the opportunity to visit the sponsorsite and evaluate existing systems to the same degree. How-ever, without having a proactive approach to gaining theinformation about the other participant in such an interac-tion, there will be no basis in the future to discuss anydevelopment or change in the relationship or change in theinteractions of the two electronic systems, and no basis onwhich to find the appropriate resolutions for issues in eitherarena.

With a formal process available, both sponsor and vendorwill be able to ask the appropriate questions and to state thesituation within their organization and the expectations theyhave for any work done within the relationship.

Four concept areas, qualification, validation, control, andinterchange, are critical for manual and electronic automa-tion systems and these will be addressed:

QualificationIn a manual system, the personnel are qualified according totheir education, training, experience, and their resultantability to work within a standard operating system. The workmust be correct with respect to scientific and technological aswell as regulatory requirements. The qualification of manualsystems can be performed by the review of training records,review for consistency and completeness of the SOP system,and review for completeness in documentation of changes orof investigation and resolution of deviations or failures.

Electronic systems can be used for technical (facilities and

Page 67: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 2

4 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

equipment) control and for record and associated data gen-eration and control. Qualification of these systems includesreview of the technical capabilities and connectivity of thehardware, suitability of the software to perform the entry,manipulations, and storage of information (whether numericor text), and review of the audit trail functionality. In addi-tion, one must take the human interactions with the systeminto account. These interactions will be additional to the onesutilized in any manual system and must be specificallyqualified for the electronic system. This evaluation can followthe same approach as stated for a manual system, but mustbe designed specifically for the electronic system being re-viewed.

ValidationIn a manual system, validation occurs in a two step process.First, the assessment for competence of the personnel whowill act within the system and the completeness of theirtraining is performed and then the conceptual evaluation forlogical consistency and completeness of the linking proce-dural system is performed. These are some of the accountabletasks within the purview of the management for the qualitysystem. Careful review and judgment of these personnel inthe beginning, and continual review and assessment must bedone in order to satisfy this validation requirement. For amanual system, this is documented by the signature of theresponsible individual on specific documents (e.g., SOP ap-proval, training records, etc.).

Validation of the electronic system must be performedaccording to written protocols for the performance of thehardware, performance of the software, and review that thedata and associated audit trails are complete. Such protocolsmust be pre-approved for completeness prior to execution andthen the executed results and conclusions also must bereviewed and approved. These exercises add to the evalua-tion and validation of the operations of the personnel andtheir capabilities to perform tasks associated with the elec-tronic system.

ControlIn a manual system, control is usually dependent on theclarity of the written Standard Operating Procedures (SOPs),the completeness of this procedural system, and on theintegrity of the qualified personnel who are performing thetasks. This is documented by a signature. For critical steps,the control is further verified by a second signature, whichverifies that the task was performed correctly and approvedby the second signature.

In electronic systems, this control is managed by theelectronic algorithm which requires steps to occur in a certainsequence, and prevents further progress of the process with-out the appropriate input of the required signatures. Thislimitation on access to the system by personnel is one of themost important control features from a management per-spective. This control is documented in the audit trail withinthe system. This is the reason that great emphasis must beplaced on establishing and maintaining the appropriate

audit trail in every electronic system.

InterchangeIn a manual system, exchange of information can be verbal orwritten, and can occur through telephone, fax, or computer e-mail. However, the only relevant information is that which iscontained in the accepted format in the formal paper docu-mentation system of the company with final approval signa-tures.

Electronic exchange must occur over very secure elec-tronic systems only, and in systems which have the internalcapability to monitor all exchanges, capability to allow inputonly by approved individuals according to allowed accessprocedures and identification algorithms, and to record alltransactions in the audit trail. As mentioned above, theseexchange-of-information interactions must occur accordingto certain rules, whereby the personnel who may enter orapprove data are known and documented by their systemname and access codes within the system. All entries, ar-chives, and changes are challenged for correctness and as-sured for completeness by the system.

Automation Systems -Similarities and Differences

There are advantages when the sponsor and vendor both areusing the same electronic automation system. In this case,each party already has an understanding of the structure,working algorithms, capabilities, and limitations of the sys-tem. And therefore the evaluation of qualification, valida-tion, control, and interchange can occur much more readily.On the other hand, an electronic system can be configuredand customized for utilization within any organization, andtherefore, each party will need to understand the specificconfiguration and any customizations that have been done bythe other party when performing the full evaluation.

When the sponsor and vendor each have a different com-puter system, additional time and effort will be needed tocomprehend the system structure, working algorithms, capa-bilities, and limitations. In this case, the documentationassociated with the installation and validation of any systemwill be very helpful. One of the best plans for dealing withaudits by regulatory authorities is to have a descriptive bookof information on each electronic system, similar to the SiteValidation Master Plan for facilities and operations, that hasa clear presentation of what the system is, how it works, whooversees its use, how the organization functions interactthrough it and links to all of the pertinent qualification,validation, and change information. This same book of infor-mation can be a very strong basis for the other party in asponsor/vendor relationship to have an understanding of itselectronic system.

Optimal Interactions - Sponsor/VendorOptimizing the interactions between the sponsor and vendordepends on the personnel and the systems that have beendeveloped in each site to support an interaction. Thesesystems should include the development, approval, execu-

Page 68: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 2

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 5©Copyright ISPE 2004

tion, archiving, and retrieval of electronic information anddata to ensure accountability, responsibility, and consis-tency. These systems will be linked to, but separate from, theexisting GMP systems for manual records. Careful consider-ation must be given to establishing the connections forlinking the in-house systems with the external systems.Some best practices include:

• clear understanding by each party of the capabilities of itssystem, desired information that needs to be exchanged,and the connection points for each category of informationand data

• well written quality agreement, mutually accepted se-crecy agreement to protect intellectual property, and con-tract

• clearly defined data and information structure in order toavoid ambiguity and to ensure easy incorporation intodata bases and final reports

• strong interactions between the two quality units to en-sure constant monitoring of systems and processes fordata integrity and security in each site

• constant checking to ensure that the process of electronicinformation exchange is still the most efficient and effec-tive possible

• periodic evaluation for update in hardware and softwareincluding periodic revalidations

• Agreement stating who will analyze the data, with whichalgorithms, who will review and approve, who will incor-porate into reports, and how, and for how long the data willbe archived

• establish well trained, knowledgeable, and experiencedpoint persons in each organization

SummaryAll companies will rely on outsourced support of clinical trialsupplies preparation at some time. And the most effectiveinteractions will occur when the best electronic systemspractices are employed by each party. Such interactions willrely on clear identification of all interaction points andclearly defined procedures and control for all tasks. Somepoints to consider and current best practices have beenpresented here for each of these aspects.

References1. US 21 Code of Federal Regulations 211.25(b) and (c).2. EC Directive (91/412/EEC) Article 7 (2) and (3).3. EC Guide to Good Manufacturing Practice for Medicinal

Products and Active Pharmaceutical Ingredients, 4th Edi-tion, Chapter 1, Quality Management, Principle.

4. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th Edi-tion, Chapter 7, Paragraphs 7.6 and 7.12.

5. US 21 Code of Federal Regulations 11.6. EC Guide to Good Manufacturing Practice for Medicinal

Products and Active Pharmaceutical Ingredients, 4th Edi-tion, Annex 11.

7. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th Edi-tion, Chapter 4, Paragraph 4.9.

8. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th Edi-tion, Chapter 7.

9. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th Edi-tion, Annex 13, Point 37.

10. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th Edi-tion, Annex 16.

11. US 21 Code of Federal Regulations 58.35 and US 21 Codeof Federal Regulations 820.

12. US 21 Code of Federal Regulations 210.1, 210.2, 210.3(10),and 211.22.

13. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th Edi-tion, Chapter 1.

About the AuthorCharles F. Carney has nearly 37 years ofexperience in pharmaceutical drug productdevelopment, manufacturing, and analyti-cal control in two major international firms,Wyeth (formerly Ayerst Laboratories) andBoehringer Ingelheim. For the last 15 yearsof this period, he was the associate director,investigational supplies, in the Connecticut

R&D Center of Boehringer Ingelheim. He is now a consultantfor Seraphim Life Sciences Consulting, LLC. His responsi-bilities have included performing and managing the pharma-ceutical, analytical, and GMP compliance aspects of drugproduct development, including the outsourcing, validation,and technology transfer aspects, and the quality control andfacilities design for commercial drug products. He is a mem-ber of ISPE, ACS, and AAAS. He has published and presentedpapers extensively in the areas of preformulation, formula-tion, and control of manufacturing operations for clinicaldrug products. He also has published and presented conceptsin the areas of outsourcing of manufacturing and control andfor optimizing international operations. He is a co-editor ofthe second edition of “Drug Products for Clinical Trials”which will publish in 2004 (Marcel Dekker, Inc.). Carney canbe contacted by tel: 1-203/426-1860 or by e-mail:[email protected].

Seraphim Life Sciences Consulting LLC, 25 Head ofMeadow Rd., Newtown, CT 06470.

Page 69: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 1©Copyright ISPE 2004

External Sourcing of Clinical TrialMaterialsPart 1: Process and Points to Considerby Charles F. Carney

This articledescribes anobjectiveprocess andpoints toconsider informulating theplan anddecision tooutsource anyof the clinicaltrial suppliesproduction,distribution, andreconciliationsteps.

Introduction

The pharmaceutical industry continuesto look for ways to optimize performance.Many companies in this industry havebegun to re-evaluate what tasks they

will consider core competence for performanceinternally and what tasks are not core compe-tence and should be performed externally. Somehave even proposed that pharmaceutical com-panies would perform best if they only per-formed research tasks and marketing tasks,and outsourced all of the developmental tasks.

One of the developmental tasks that is oftenconsidered for external sourcing is the prepa-ration of clinical trial supplies. Today, manycompanies source some or all of the manufac-turing, packaging, labeling, holding, distribu-tion, and ultimate reconciliation and destruc-tion of some clinical trial supplies through theutilization of contract firms. When queriedabout this, many reasons are given for theutilization of contracted capacities, includinglack of labor or machine capacity, lack of theappropriate technology, or lack of expertise inthe particular requirements for completing thetask. For many, the decision to outsource seemsto be made subjectively, using past practiceand experiences as the guides with each casebeing independently considered. Others maybe looking for an objective process. Therefore, itseems appropriate at this time to attempt tocodify the logical bases for external sourcing ofclinical supplies preparations. This will pro-vide an objective process and decision map forthose who may not yet have developed a sys-tematic decision making process for theseoutsourcing decisions.

Organization and Decision MakingEvery organization needs to develop the strate-

gic statements relative to outsourcing the prepa-ration of clinical trial materials. This is neces-sary to ensure that the best leverage is gainedfrom any external interactions. It also ensuresthat all internal functions are agreed withrespect to dealing with the relationship withanother company. If outsourcing does not fitinto the corporate strategic plan, the clinicaltrials supplies group should not even considerthis possibility when planning for the execu-tion of their work assignments. On the otherhand, if outsourcing does fit into the corporatestrategic plan, everyone in the organizationshould know this and be ready and willing tosupport the decision. This support will includethe development of the interaction relationshipand modification of any systems necessary toget the appropriate quality and quantity ofwork performed using external capacity. Thesefunctions will include all of the relevant func-tions in R&D (for example, pharmaceutics,analytical, quality assurance, project manage-ment) as well as the support functions in thecorporation (for example, legal, purchasing,shipping and receiving).

The procedural system will need to be ex-panded to include those procedures necessaryfor the qualification and controlling of contractvendors. And this will necessitate additionaltraining modules as well. The organization willneed to determine who will make the decision,ultimately, for the outsourcing of work. Will theclinical trial group manager make this decisionor will it be made by a team, by a team leader,an assigned outsourcing point-person, purchas-ing, or someone in the medical research group?The decision should usually be taken as aconsensus view of the stakeholders, which ispredicated to making the delivery to the clini-cal trial at the time needed by the customer.

Reprinted from The Official Journal of ISPE

PHARMACEUTICAL ENGINEERING® September/October 2004, Vol. 24 No. 5

Page 70: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

2 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

The stakeholders in this case include all of the R&D functions(pharmaceutics, analytical, quality assurance) and the per-sonnel in the medical department who will manage theexecution and performance of the trial.

Outsourcing of work always puts an extra layer of burdenon the sponsor. This burden affects each function at differenttimes and in different ways. Providing the formulation andprocess for manufacturing of product at a contract site willhave different demands and constraints than providing ana-lytical test methods to the vendor or performing testing onsamples coming from the vendor. Similarly, the burden onthe quality assurance group will increase in their need toreview systems and personnel performance both at the homesite and at the vendor site. The shipping and receivingpersonnel also will have some constraints and demands onthem for interacting with another group in another company.All of the interactions pathways, point persons for the inter-actions, and ways of interacting need to be developed andimplemented prior to engaging in the first outsource project.

Traditionally, cost has not been an issue for clinical sup-plies operations, and therefore, the outsourcing of work wasdriven by the need to meet the time-line rather than the costof the work. This may be changing today as pharmaceuticalcompanies become more and more cost conscious in all areasof the business, including R&D operations. In addition, thetrend today toward increased work load without increasinghead count needs to be considered. Internal capacity isrequired to establish, oversee, and evaluate a relationship forusing external capacity. This fact must not be overlooked inthe development, implementation, and execution of anoutsourcing program.

Multinational corporations will need to consider the im-pact on their outsourcing strategy when they have more thanone clinical supplies group, each one in a different country,and even when the corporation is big enough to have morethan one clinical supplies group in the same country. Suchcompanies may want to have harmonized systems foroutsourcing by all of the units in order to take best advantageof a relationship with vendors and to be able to negotiate thebest prices with the vendors. The majority of clinical trialstoday are executed in multiple regulatory regions and mul-tiple countries within each region. Also, each of the clinicalsupplies preparation vendors is able to work in all of theseregions. Therefore, it may be beneficial to develop the work-ing relationship on a corporate, rather than working unit,basis with each vendor. Harmonization of the work practiceswithin the corporation for the various clinical supplies groupswill pay dividends when the time comes to outsource a majorclinical trial being performed in each of the regulatory re-gions in the world. The best utilization of internal capacityand the most appropriate utilization of external capacity canbe assured using this approach.

This approach will work best when the interaction of thevarious personnel in the medical research disciplines in thecorporation, in the various countries, also are harmonized.Consistency in the ways that requests are made, informationis exchanged, and in the management of all changes between

the medical and clinical supplies personnel will allow optimalperformance of the execution work by the clinical suppliespersonnel.

The medical customer will always want supplies “on-demand.” In addition, changes will always be needed in thepreparation of supplies for each trial today. Protocols arecomplicated and guarantees for completion of all necessaryregulatory hurdles, prior to the initiation date, are not pos-sible. These changes-in-thinking of medical groups placesgreat demands on the logistic flexibility and “magic-working”of CTM groups, including the work of their contract partners.

Currently, the costs for contracting of CTM preparationsare reported to be greater in the EU than in the US. Thesecosts include both the relative higher costs charged by thecontract firms as well as the interaction costs for moving thematerials between sponsor and vendor, because often thevendors are not in the same country as the sponsor. The UKand Ireland have the largest population of full supportvendors which means that the mainland countries mustexport drugs to them for work. The interactions costs includeimportation and exportation fees, and less tangible costs forinteractions in a “foreign language.”

Special Points for Sponsorto Consider for any Project

With the strategy established, the personnel responsible foroutsourcing need to consider how best to leverage the utiliza-tion of external capacity. This planning should include a longterm view of the needs for outsourcing for specific trials or forwhole programs.

For each trial, planning must account for a one time supplyfor the whole trial or whether initial supplies and follow-upsupplies are needed for the successful execution of the trial.If the trial is a long term trial, then the capacity needs can bebest supplied by an external partner, unless the sponsor isequipped with the necessary space and labor capacity tohandle both new programs and ongoing programs. It seems,from casual conversations, that few sponsors are adequatelystaffed today or have adequate space to handle all of the shortand long term programs in their portfolio. This is particularlytrue with respect to the increasing need for fast and flexibleexecution of the early phase program required today for proofof principle or final choice of compound to take to full devel-opment toward commercialization.

For the case that the sponsor is planning for multiple trialsin a project, or perhaps for all of the trials in that project, thesponsor needs to consider whether there are any specialhandling needs, for example, refrigeration for proteins, extraspace for individually packaged inhalation products, or shipon demand scenarios in the protocol to name a few. It isusually best to execute all of the packaging and labelingefforts in one site because of the efforts needed to establishand maintain the special conditions. In these cases, whenoutsourcing is necessary, a good decision is to plan for theoutsourcing of all trials for that program. This may allow thebest price negotiation and the most cost effective systemdevelopment to accommodate the special needs for the mate-

Page 71: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 3©Copyright ISPE 2004

rials for the trials.New or special designs for trial protocols provide an

opportunity to learn, but also may prove to be taxing to theclinical supplies group that is already working at full capac-ity. For example, when a new randomization model is needed,it is necessary to ensure that adequate validation of that newalgorithm has been performed if this algorithm will be devel-oped and implemented by the vendor. The development andexecution of this validation should be a special considerationin the request for proposal sent by the sponsor to the vendor.This also will require that the sponsor perform a special re-evaluation of the vendor to ensure that it can be doneadequately, both from the technology perspective and fromthe personnel expertise perspective. Therefore, part of thestrategic thinking within the sponsor group should be anevaluation whether all randomizations will be supplied bythe sponsor or whether the sponsor will rely on the vendor toperform these, and to manufacture, package, and label ac-cordingly.

In some cases, special packaging configurations are neededto ensure subject compliance with the drug administrationrequirements of the protocol. Subject and investigator com-pliance with a protocol can oftentimes be enhanced by themanner in which the goods are packaged and labeled. Suchspecial considerations as colored labels, patient box size,configuration of goods within a patient box, sizes of primarycontainers, and numbers of dosage units per container canoftentimes make a significant difference for ensuring compli-ance and for assisting in the reconciliations and accountabili-ties of drugs at the sites. On the other hand, it’s very easy tooverestimate the gain for compliance or accountability bysuch customizations. The personnel who are responsible formaking the clinical supplies configuration decisions in thesponsor site need to have an objective view to support anddefend such requests from the customer especially whenoutsourcing the work. The vendor also should be ready toassist in finding creative compliance-enhancing packagingand labeling which is also efficient and effective in theutilization of labor and materials.

Planning for the use of outsourced capacity should antici-pate the need for changes at a later date. Both the sponsor andthe vendor should anticipate that something may changeduring execution of the preparation or during execution of thetrial. A mechanism for communication and for addressing therequest for change and the change control should be estab-lished and maintained between the sponsor and vendor.

Similarly, planning for special distribution considerationssuch as stratified distribution according to some algorithmfor differences in sex, race, or geographical location of thecenter should occur. The need for the special distributionneeds to be understood by the clinical supplies group andcommunicated to the vendor. If any special computer pro-gramming is necessary, all of the qualification and validationrequirements for computers need to be considered and imple-mented.

In some cases, an Interactive Voice Response System(IVRS) will be needed in order to manage a complicated

distribution or the control and distribution of limited drugsupplies. These systems are expensive to define and to estab-lish, to train the personnel at the sites, to provide the correctlinkage between the supply labels and the controlling sys-tem, and to ensure that the correct information is collectedand managed according to the requirements specified in theprotocol. Such systems are not to be used with every trialbecause of the costs involved, but can improve the collectionand collation of information from complex trials. The seconduse of these systems can be for the conservation of drugproducts which are in short supply or to limit the amounts ofmaterials distributed for controlled drugs.

If the use is for controlled drugs, the sponsor needs toensure that the vendor has the appropriate physical plantand licenses. State and federal DEA licenses are required forthe specific scheduled compounds and products that will behandled. In addition, specific physical access and monitoringcontrols need to be in place, as well as very detailed SOPs andpersonnel training, specifying the manner and methods formanaging, recording, and destroying the drugs. The specificregulatory requirements of each country in which the con-trolled drugs will be studied must be considered for importa-tion, distribution, accountability, and reconciliation. Chainof custody documentation and training of personnel in boththe sponsor and vendor sites are important aspects for thesuccessful management of scheduled drug products

The importation and exportation requirements for thetrial and the abilities of the vendor to supply support of abonded broker and good understanding of requirements andlimitations need to be ensured continually because the rulesand requirements change routinely.

Establishing and MaintainingRelationships with Vendors

The sponsor/vendor relationship is the cornerstone of eachoutsourcing success. A fully elaborated concept for the rela-tionship and the efforts to develop this relationship proactivelyare necessary to ensure successful outsourcing of work. Bothsponsor and vendor must have a similar quality position anda similar vision of success in order for the interactions todeliver results. The key points for establishing and maintain-ing the relationship include formal qualification of the ven-dor by the sponsor, development and implementation of theinteraction procedures and responsibility assignments, utili-zation of a formal, detailed requesting and proposal develop-ment/acceptance routine, and a communication system thatis objective and covers all interactions.

The comfort level and continuing allocation of work to avendor depends on the initial impressions about capabilitiesof the vendor and the ongoing experience with the vendor.These are best evaluated by a formal process for identifyingpossible vendors and qualifying each for technical, quality,and business performance attributes. The initial impressionof technical competence by the clinical supplies group and ofquality compliance by the QA group will be lasting impres-sions that will influence all future interactions. A formalapproach should be utilized for these evaluations. This ap-

Page 72: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

4 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

proach will start with the development of an inspectioncheck-list to ensure that all points are evaluated. This isimportant whether the evaluation is for technical, quality, orbusiness competence.

Important aspects for the preparation of clinical supplieswill include the depth and breadth of technical competence inthe vendor organization which can ensure a rapid under-standing of the job to be done and recommendations for thebest contemporary scientifically and technologically soundways to do it. Similarly, the quality organization must be up-to-date on all national and international regulatory require-ments and expectations. It also must have the depth andbreadth of contemporary quality assurance concepts to bethoroughly in-control while always looking for the flexibilityto handle the ever changing landscape of clinical suppliespreparations.

These evaluations should explore the most recent experi-ences, including both positive and negative experiences.Some negative aspects that require scrutiny include anyincrease in error frequency, complaints from the field forgoods produced by the vendor, poor understanding of thescience or technology that they have available in their shop,lack of ability to interpret protocol correctly, lack of ability toprovide innovative and cost effective manufacturing, packag-ing, and/or labeling concepts for the protocol, and inability toexpress their corporate philosophy or expectations for theinteractions.

The sponsor/vendor relationship needs to be maintainedat the highest level of satisfaction by constant communica-tion, honesty, openness, and teamwork. The importance ofstrength in technical and regulatory understanding on aninternational basis cannot be overemphasized for the successof this relationship.

For clinical supplies, quality and time to delivery usuallytake precedence over cost. However, cost has become a fargreater factor for most sponsors today and every sponsorshould look for some preferred pricing schedule from a vendorand also for any reasons for giving work preferentially to thatvendor even when the price is higher than that offered byothers. In some cases, there exist intangible points whichmake one relationship work better than another. Theseintangibles include personality compatibility, demonstra-tion of dedication to the desires of that sponsor, responsive-ness, and inventiveness. In most cases, these attributescannot be measured in cash terms, but become the decisivefactors in the successes achieved through the interactions.

The Quality structure of the vendor, particularly if thatvendor has been associated with any kind of regulatory actionor denial by the FDA or any other agency, is extremelycritical. The sponsor should look for those vendors whoconstantly reinforce the understanding that they performonly value added work according to the most contemporaryinterpretation and application of the cGMP on a worldwidebasis.

A procedural basis for the interactions should be estab-lished between the sponsor and vendor. This will mean thatthe SOP system of the sponsor will need to be expanded to

include those activities with the external organizations. Howthe sponsor and vendor interact is critical, especially whenprotocols change or negative issues arise. The best approachis for the sponsor and vendor to develop and agree on theinteraction model that will be employed. Otherwise, miscom-munication and misunderstanding may impact the deliveryof very important trial supplies.

Technical/Regulatory understanding by the vendor is cru-cial, and the sponsor should seek out the vendor with the bestunderstanding and who turns this understanding into prac-tice. The technical understanding must include formulationdesign, manufacturing techniques and process flow, packag-ing materials science, closure integrity factors, equipmentfunction and its impact on the manufacturing, packaging orlabeling, of product, and computer design, qualification, andvalidation to name a few. Regulatory understanding mustinclude the cGMP requirements of the regulatory regions inwhich the clinical trials will be performed. At a minimum, thevendor must have full competence in the requirements of theUS and EU for drug products and biological drug products,and devices, safety standards and requirements particularlyfor potent or potent/hazardous products, regulations for trans-portation, importation/exportation requirements and restric-tions, and intellectual property regulations including patentand trade dress issues.

The connection between the quality systems of sponsorand vendor must be clear and include a thorough understand-ing of roles and responsibilities. This should take into accountthe requirements for decision making according to role of theQuality Control Unit of 21 CFR1 and also of the qualitymanagement system, the Qualified Person, and the specialapplication of the GMP for clinical supplies in the EuropeanUnion.2

The quality management of all materials, whether forquality control testing or for quality assurance evaluation,review and approval, must be clearly specified for the spon-sor/vendor relationship. Responsibilities for the quality con-trol sampling and testing for starting materials, whetheractives or excipients, in-process materials, packaged andlabeled supplies, packaging materials, environmental vali-dation samples, and purified water, to name a few, must beclearly defined. It should be clear which testing will be doneby the sponsor and which will be done by the vendor. Under-standings and agreements concerning the standards andacceptance criteria for this testing must be established be-tween the sponsor and vendor. Subcontracting of analyticalwork, or any other work, by the vendor must be understoodand agreed by the sponsor. And the review, approval, anddisposition process and assigned responsibilities also mustbe clearly specified. All of this should be summarized in awritten form and approved by both organizations.

A formal process should be developed for requesting workby the sponsor and the estimates of performing the work bythe vendor. The sponsor’s Request For Proposal (RFP) shouldinclude a detailed summary of all that will be needed, whowill supply materials, the requested timing for completion, acopy of the trial protocol for reference, and any other specific

Page 73: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 5©Copyright ISPE 2004

information necessary for the completion of the work. Itshould include all the necessary information in an objectiveand clear layout. Similarly, the vendor’s proposal shouldaddress all points and should provide the confirmation ofacceptance of responsibilities and agreement for providingthe desired deliverables and for meeting the desired timelines. Costs for the work should be clearly and explicitlydetailed in order for the sponsor to have a thorough under-standing of the work that the vendor will provide and toensure that the sponsor will receive the work that is needed.

Finally, and perhaps most importantly, a communicationsystem must be established. This will work best if pointpersonnel are assigned in the sponsor and vendor organiza-tion to ensure the most efficient and effective transmission ofknowledge, desires, expectations, data, and decisions withineach organization. This system also must include a processfor resolution of conflict between the two organizations.While the desire in both companies will be for “no conflict”processes, nevertheless, on occasion, errors, delays, changes,and mishaps may interfere with any smooth process whichhas been envisioned. Objective and speedy resolution of theseinterruptions in the process must occur in order to maintainthe cost effectiveness of the interactions.

An assumption that underlies all of these statements isthat a contract between the sponsor and vendor will benegotiated and executed prior to any work being performed.Discussion of the development and establishment of terms ofa contract would require a separate article altogether, andtherefore, will not be further elaborated here. For the purposeof this discussion, realize that a contract is required and theelements of such a contract are clearly delineated in theEuropean regulations3 and understood through the concept ofcurrent best practices in the US regulations.

Decision PathwayOnce the sponsor/vendor relationship has been established,the sponsor which utilizes both in-house capacity and exter-nal capacity will need a logical, decision making pathway.Such a decision tree will ensure that objective decisionmaking occurs for each case of outsourced work. One suchdecision tree could be imagined and is laid out below. Otherscould be imagined or this one could be modified to meet theexact requirements and organization of any vendor. It ispresented here as a framework within which specific detailscan be added by any sponsor.

A. Primary considerations for a decision: available labor,technology, space for staging/storage, special distributionrequirements, IVR requirements, uniqueness of the proto-col, special temperature or relative humidity handling,other?1. If labor, technology, or space are not available at the

time required for delivery of supplies, outsource.a. If temporary contracted labor is available and tech-

nology and space are not concerns, perform in-house.2. If experience with preparation for the special distribu-

tion requirements exists, perform in-house, otherwise

seek external help.3. If IVR is required, exists in-house and available, uti-

lize, otherwise outsource.a. For outsource, ensure that the IVR and clinical

supplies preparation vendors are compatible andwork together well.

b. For in-house or outsource operations, education maybe needed for the Medical customer and all interac-tions with internal groups (Medical, Formulation,Purchasing, Analytical, Traffic, Regulatory Affairs)must be established and tested.

c. Develop validation plan for the IVR system andcomplete execution prior to the start of the trial.

4. If special randomization, stratification, or other out ofthe ordinary packaging or labeling are needed by theprotocol, and these can be accommodated in-house,then do so, otherwise seek external help.a. Ensure all US4 and EU5 computer qualification and

validation requirements are taken into account.5. If temperature or relative humidity, or special configu-

ration shipping is needed (for example, cold shipments,or containers must be in upright position), ensure thatshipping conditions and materials required have beenconfirmed by the shipping test.

6. Evaluate for any special import requirements (for ex-ample, USDA import permit for animal derived orbiological materials, IND in place for API or drugproducts, patent or trade dress issues for comparatordrugs); and for export constraints (for example, perfor-mance of the clinical protocol only in Listed countries oralso in non-Listed countries, SGA prequalification forshipments to Philippines, import license availabilitiesin the countries to which the products will be sentincluding special “third country audit” requirementsfor Germany, special documentation and permits forscheduled drugs, and CTA in those countries whichrequire it).6

7. The Traffic group must maintain an awareness ofspecial handling needs and the organization shouldensure that bonded import brokers exist in all countrieswith sufficient capital to pay duties and expertise tomanage the importation efficiently.

B. After the decision to outsource has been made: ensure thatthe final quality of the deliverables has been specified,determine the time needed to prepare supplies, determinethe desired time of delivery, estimate the costs if the workwere to be performed in-house, estimate the degree offlexibility for changes that may exist in the project, andchoose to submit the request for proposal to the vendorwith the greatest likelihood to meet the expectations forthe project. If a competitive bidding process is required inthe organization, the request will need to be submitted tothe minimum number of appropriate vendors, as requiredby the organization.1. Include the requirements and clear definition for what

work is requested of the vendor in the Request For

Page 74: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

6 PHARMACEUTICAL ENGINEERING Supplement SEPTEMBER/OCTOBER 2004 ©Copyright ISPE 2004

Proposal (RFP), including all of the relevant informa-tion gathered in A.

2. Determine whether the vendor will be chosen by com-petitive bid (cost) or by previous experience with aparticular vendor in whom confidence exists because ofproven record (quality) or by the evaluation for shortestperiod for delivery of goods (time).a. Traditionally, the decision has been determined by

which vendor will provide the best Quality andshortest Time to Delivery over lowest Cost in mak-ing the sourcing decision. Recently, there has been amuch stronger push toward competitive biddingbetween vendors to get the best price. While cost iscritical, it is important that purchasing groups un-derstand that the cost of preparing clinical trialsupplies is a very small part of the overall costs ofthat trial, or the program for which that trial is beingperformed. It may prove beneficial to spend sometime educating the purchasing agents that clinicaltrial materials are not commodities, and that manyother intangible aspects need to be considered in thedecision. These include, but are not limited to:i. technical understanding by the vendor for the

protocol needs and their own ability to providethe supplies to meet the protocol, in the timeframe

ii. flexibility and willingness of the vendor to accom-modate changes necessary during execution ofthe preparation because of new information re-ceived from the medical customer

iii.ability of the vendor to overcome problems, er-rors, delays in receipt of materials, suddenlyrecognized deficiencies in materials supplied bysponsor, and to work with the sponsor to providenecessary investigations, solutions, and docu-mentation required for adequate record keeping,control, and final decisions or approvals

iv. honesty and openness of the vendor in recogniz-ing own deficiencies, lack of understanding, ormistakes, and the honesty of open communica-tions to resolve all issues objectively

v. willingness of the vendor to ask questions forclarification and understanding

b. If bidding is used, it must be objectively based. All ofthe vendors to whom the RFP is issued must haveequal capabilities and have been equally evaluatedaccording to their technical competence and regula-tory compliance position. Not all vendors have thesame strengths and capabilities, as outlined above.i. Prior decision should be made as to the accep-

tance criteria for a bid (for example, take thelowest, or take the middle, or…).

ii. Turn around time for the proposal and bid and thecompleteness of the proposal to address the spe-cific points in the RFP should be considered.

iii.Preferred partner or guaranteed work pricing ispreferred for the preparation of clinical trial ma-

terials over simple straight bid process becausethese supplies are not commodities, but ratherare prepared uniquely for each protocol.

Some additional considerations for the size of the work beingoutsourced, whether to outsource only the specific trial or allof the trials for the program, and how many different vendorsto manage at one time should occur in order to evaluate, plan,and execute the outsourcing decisions. The most cost effectiveapproach is combining work at the vendor according to theproject or according to the pattern of complexity of a series ofprotocols for related programs, perhaps by dosage form typeor special packaging/labeling needs. This should take thestrengths and special capabilities of the vendor into account.It’s best to ask the vendor, during the evaluation, to statestrengths and niche capabilities. Some scale advantages andconservation of concept development time can occur whenlike jobs are bundled at a vendor site.

The sponsor needs to have a concept for what it will do in-house and what is best done externally. For example, thesponsor with limited staging and warehousing space needs toconsider doing only small preparation jobs (Phase I and IItrials) in-house and contracting the larger (Phase III and IV)jobs.

And the sponsor also should think carefully about longterm protocols, for example three to five year duration proto-cols requiring significant re-supply and continuation supplyefforts. This decision making will depend on whether thesponsor decides to spend time ensuring that the new projectsget the best in-house attention, when there is the greatestneed for attention to ensure the project will go forward, orwhether to attend to the near market projects for which thereis greater certainty of success, and therefore, more surety ofincome. Many clinical supply groups focus on the smaller,early phase, projects, in-house and contract out the larger,later phase, less risky, but more capacity and time consum-ing, jobs. However, each sponsor needs to make this assess-ment according to its strategic and tactical thinking.

Distribution ConsiderationsDistribution of clinical supplies is complex and can be besttreated in a separate, devoted, article. Only some brief com-ments will be made here with respect to the outsourcing ofdistribution activities for distribution from the US.

Distribution from the US is limited by the complicatedexport laws. These have relaxed in the last few years since therenovation of the law in 1996. However, with the new con-cerns about terrorism some congressional members are talk-ing about tightening both the importation and exportationcontrols. The current law can be used to best advantage byperforming all trials according to the requirements of the USIND,7 by performing foreign site trials in Tier One (listed)countries,8 and by applying early for authorization to ship tothe non-Tier One (non-listed) countries.9 There seems to besome relaxation for transshipment of CTM through a Tier-One country to a non-Tier-One country if the Principle Inves-tigator in the Tier-One country takes accountability to ensure

Page 75: Reprinted from The Official Journal of ISPE …based on the Resistance vs. Temperature tolerances in ASTM E1137 or IEC 60751, which at 100°C are ±0.3°C for the Grade A or Class

External Sourcing - Part 1

SEPTEMBER/OCTOBER 2004 PHARMACEUTICAL ENGINEERING Supplement 7©Copyright ISPE 2004

that the trial will be conducted according to the principles andpractices in place in the Tier-One country. This developmentneeds to be watched carefully and this first impression needsto be validated before transshipments of this nature becomeroutine. If distribution will be managed by the vendor, a veryclose collaboration between sponsor and vendor will be neededto ensure that both organizations understand and complywith all of the requirements.

SummaryUtilization of contracted resources for the preparation ofclinical trial supplies has increased significantly over the lastdecade. An objective process, which fits into the tacticalrequirements of the sponsor company required to meet itsstrategic goals for success, will be the best approach. Such anobjective process has been outlined here with some emphasison the points to consider for working in the internationalarena.

References1. 21 Code of Federal Regulations 211.22.2. EC Guide to Good Manufacturing Practice for Medicinal

Products and Active Pharmaceutical Ingredients, 4th. Ed.,Chapter 1, Annex 13, and Annex 16.

3. EC Guide to Good Manufacturing Practice for MedicinalProducts and Active Pharmaceutical Ingredients, 4th. Ed.,Chapter 7.

4. 21 Code of Federal Regulations 211.68 and 11.1.5. EC Guide to Good Manufacturing Practice for Medicinal

Products and Active Pharmaceutical Ingredients, 4th. Ed.,Annex 11.

6. The importation and exportation requirements in eachcountry are complicated and subject to change. Completeelaboration of these therefore cannot be provided in thisshort article. An appropriate tactical approach for main-taining the knowledge base may be to research specificquestions through Web-based sites, such as www.fda.govand www.usda.gov. A database for governmental regula-tory contact information has been established atwww.ispe.org which can prove useful for questions relat-ing to clinical supplies. Another organization maintains auseful file of direct links to the various, pharmaceuticallyrelevant, governmental Web sites at www.pharmedassociates.com/links.asp.

7. 21 Code of Federal Regulations 312.110(b)(1).8. 21 Code of Federal Regulations 312.110(b)(4).9. 21 Code of Federal Regulations 312.110(b)(2).

See External Sourcing - Part 2 for About the Author.