Process validation in USA

Cleaning Validation
Сообщения: 45
Зарегистрирован: 16 мар 2012 10:31

Process validation in USA

Сообщение Dima » 05 апр 2012 15:11

Process validation is establishing documented evidence which provides a high degree of assurance that a specific process (such as the manufacture of pharmaceutical dosage forms) will consistently produce a product meet¬ing its predetermined specifications and quality characteristics.
According to the FDA, assurance of product quality is derived from care¬ful and systemic attention to a number of important factors, including: selection of quality components and materials, adequate product and process design, and (statistical) control of the process through in-process and end-product testing.
Thus, it is through careful design (qualification) and validation of both the process and its control systems that a high degree of confidence can be estab¬lished that all individual manufactured units of a given batch or succession of batches that meet specifications will be acceptable.
According to the FDA's Current Good Manufacturing Practices (CGMPs) 21CFR211.110 a:
Control procedures shall be established to monitor output and to validate performance of the manufacturing processes that may be responsible for causing variability in the characteristics of in-process material and the drug product.
The first four items listed above are directly related to the manufacture and validation of solid dosage forms. Items 1 and 3 are normally associated with variability in the manufacturing process, while items 2 and 4 are usually influenced by the selection of the ingredients in the product formulation. With respect to content uniformity and unit potency control (item 3), adequacy of mixing to assure uniformity and homogeneity is considered a high-priority con¬cern.
Conventional quality control procedures for finished product testing en¬compass three basic steps:
1. Establishment of specifications and performance characteristics
2. Selection of appropriate methodology, equipment, and instrumenta¬
tion to ensure that testing of the product meets specifications
3. Testing of the final product, using validated analytical and testing
methods to ensure that finished product meets specifications.
With the emergence of the pharmaceutical process validation concept, the fol¬lowing four additional steps have been added:
4. Qualification of the processing facility and its equipment
5. Qualification and validation of the manufacturing process through appropriate means
6. Auditing, monitoring, sampling, or challenging the key steps in the process for conformance to in-process and final product specifications
7. Revalidation when there is a significant change in either the product or its manufacturing process [3].
It has been said that there is no specific basis for requiring a separate set of process validation guidelines, since the essentials of process validation are em¬bodied within the purpose and scope of the present CGMP regulations [2]. With this in mind, the entire CGMP document, from subpart В through subpart K, may be viewed as being a set of principles applicable to the overall process of manufacturing, i.e., medical devices (21 CFR-Part 820) as well as drug prod¬ucts, and thus may be subjected, subpart by subpart, to the application of the principles of qualification, validation, verification and control, in addition to change control and revalidation, where applicable. Although not a specific requirement of current regulations, such a comprehensive approach with respect to each subpart of the CGMP document has been adopted by many drag firms. A checklist of qualification and control documentation with respect to CGMPs is provided in Table 1. A number of these topics are discussed sepa¬rately in other chapters of this book.
The FDA, under the authority of existing CGMP regulations, guidelines [1], and directives [3], considers process validation necessary because it makes good engineering sense. The basic concept, according to Mead [5], has long been

Development stage
applied in other industries, often without formal recognition that such a concept was being used. For example, the terms reliability engineering and qualification have been used in the past by the automotive and aerospace industries to repre¬sent the process validation concept.
The application of process validation should result in fewer product re¬calls and troubleshooting assignments in manufacturing operations and more technically and economically sound products and their manufacturing processes. In the old days R & D "gurus" would literally hand down the "go" sometimes overformulated product and accompanying obtuse manufacturing procedure, usually with little or no justification or rationale provided. Today, under "FDA's Preapproval Inspection (PAI) program [4] such actions are no longer accept¬able. The watchword is to provide scientifically sound justifications (including qualification and validation documentation) for everything that comes out of the pharmaceutical R & D function.

Unfortunately, there is still much confusion as to what process validation is and what constitutes process validation documentation. At the beginning of this introduction several different definitions for process validation were provided, which were taken from FDA guidelines and the CGMPs. Chapman calls process validation simply "organized, documented common sense" [6]. Others have said that "it is more than three good manufactured batches" and should represent a lifetime commitment as long as the product is in production, which is pretty much analogous to the retrospective process validation concept.
The big problem is that we use the term validation generically to cover the entire spectrum of CGMP concerns, most of which are essentially people, equipment, component, facility, methods, and procedural qualification. The spe¬cific term process validation should be reserved for the final stage(s) of the product/process development sequence. The essential or key steps or stages of a successfully completed product/process development program are presented in Table 2 [7].
The end of the sequence that has been assigned to process validation is derived from the fact that the specific exercise of process validation should never be designed to fail. Failure in carrying out the process validation assign¬ment is often the result of incomplete or faulty understanding of the process's capability, in other words, what the process can and cannot do under a given set of operational circumstances. In a well-designed, well-run overall validation program, most of the budget dollars should be spent on equipment, component, facility, methods qualification, and process demonstration, formerly called pro¬cess qualification. In such a program, the formalized final process validation

With the exception of solution products, the bulk of the work is nor¬mally carried out at 10 x batch size, which is usually the first scale-up batches in production-type equipment.
sequence provides only the necessary process validation documentation required by the regulatory authorities—in other words, the "Good Housekeeping Seal of Approval," which shows that the manufacturing process is in a state of control.
Such a strategy is consistent with the U.S. FDA's preapproval inspection program [4], wherein the applicant firm under either a New Drug Application (NDA) or an Abbreviated New Drug Application (ANDA) submission must show the necessary CGMP information and qualification data (including appro¬priate development reports), together with the formal protocol for the forthcom¬ing full-scale, formal process validation runs required prior to product launch.
Again, the term validation has both a specific meaning and a general one, depending on whether the word "process" is used. Determine during the course of your reading whether the entire concept is discussed in connection with the topic—i.e., design, characterization, optimization, qualification, validation, and/ or revalidation—or whether the author has concentrated on the specifics of the validation of a given product and/or its manufacturing process. In this way the text will take on greater meaning and clarity.

The pilot program is defined as the scale-up operations conducted subse¬quent to the product and its process leaving the development laboratory and prior to its acceptance by the full scale manufacturing unit. For the pilot program to be successful, elements of process validation must be included and completed during the developmental or pilot laboratory phase of the work.
Thus, product and process scale-up should proceed in graduated steps with elements of process validation (such as qualifications) incorporated at each stage of the piloting program [9,10].
A. Laboratory Batch
The first step in the scale-up process is the selection of a suitable preliminary formula for more critical study and testing based on certain agreed-upon initial design criteria, requirements, and/or specifications. The work is performed in the development laboratory. The formula selected is designated as the (1 x) laboratory batch. The size of the (1 x) laboratory batch is usually 3-10 kg of a solid or semisolid, 3-10 liters of a liquid, or 3000 to 10,000 units of a tablet or capsule.

3. Cost of raw materials
4. Inventory requirements for clinical and nonclinical studies
Process demonstration or process capability studies are usually started in this important second stage of the pilot program. Such capability studies consist of process ranging, process characterization, and process optimization as a prereq¬uisite to i the more formal validation program that follows later in the piloting sequence.
C. Pilot Production
The pilot-production phase may be carried out either as a shared responsibility between the development laboratories and its appropriate manufacturing coun¬terpart or as a process demonstration by a separate, designated pilot-plant or process-development function. The two organization piloting options are pre¬sented separately in Figure 1. The creation of a separate pilot-plant or process-development unit has been favored in recent years because it is ideally suited to carry out process scale-up and/or validation assignments in a timely manner. On the other hand, the joint pilot-operation option provides direct communication between the development laboratory and pharmaceutical production.

B. Laboratory Pilot Batch
After the (1 x) laboratory batch is determined to be both physically and chemi¬cally stable based on accelerated, elevated temperature testing (e.g., 1 month at 45°C or 3 months at 40°C or 40°C/80% RH), the next step in the scale-up process is the preparation of the (10 x) laboratory pilot batch. The (10 X) laboratory pilot batch represents the first replicated scale-up of the designated formula. The size of the laboratory pilot batch is usually 30-100 kg, 30-100 liters, or 30,000 to 100,000 units.
It is usually prepared in small pilot equipment within a designated CGMP-approved area of the development laboratory. The number and actual size of the laboratory pilot batches may vary in response to one or more of the following factors:

The object of the pilot-production batch is to scale the product and process by another order of magnitude (100 x) to, for example, 300-1,000 kg, 300-1,000 liters, or 300,000-1,000,000 dosage form units (tablets or capsules) in size. For most drug products this represents a full production batch in standard production equipment. If required, pharmaceutical production is capable of scal¬ing the product/process to even larger batch sizes should the product require expanded production output. If the batch size changes significantly, additional validation studies would be required. The term product/process is used, since one can't describe a product with discussing its process of manufacture and, conversely, one can't talk about a process without describing the product being manufactured.
Usually large production batch scale-up is undertaken only after product introduction. Again, the actual size of the pilot-production (100 x) batch may vary due to equipment and raw material availability. The need for additional pilot-production batches ultimately depends on the successful completion of a first pilot batch and its process validation program. Usually three successfully completed pilot-production batches are required for validation purposes.
In summary, process capability studies start in the development labora¬tories and/or during product and process development, and continue in well-defined stages until the process is validated in the pilot plant and/or pharmaceu¬tical production.

Because of resource limitation, it is not always possible to validate an entire company's product line at once. With the obvious exception that a company's most profitable products should be given a higher priority, it is advisable to draw up a list of product categories to be validated.

Process validation is done by individuals with the necessary training and experi¬ence to carry out the assignment.
The specifics of how a dedicated group, team, or committee is organized to conduct process validation assignments is beyond the scope of this introduc¬tory chapter.
Process capability is defined as the studies used to determine the critical process parameters or operating variables that influence process output and the range of numerical data for critical process parameters that result in acceptable process output. If the capability of a process is properly delineated, the process should consistently stay within the defined limits of its critical process parameters and product characteristics [12].
Process demonstration formerly called process qualification, represents the actual studies or trials conducted to show that all systems, subsystems, or unit operations of a manufacturing process perform as intended; that all critical process parameters operate within their assigned control limits; and that such studies and trials, which form the basis of process capability design and testing, are verifiable and certifiable through appropriate documentation.
The manufacturing process is briefly defined as the ways and means used to convert raw materials into a finished product. The ways and means also include people, equipment, facilities, and support systems required to operate the process in a planned and effectively managed way. All the latter functions must be qualified individually. The master plan or protocol for process capabil¬ity design and testing is presented in Table 5.
A simple flow chart should be provided to show the logistical sequence of unit operations during product/process manufacture. A typical flow chart used in the manufacture of a tablet dosage form by the wet granulation method is presented in Figure 2.

The modern approach, based on Japanese quality engineering [15], is the pursuit of "zero defects" by applying tighter control over process variability (meeting a so-called 6 sigma standard). Most pharmaceuti¬cal products and their manufacturing processes in the United States today, with the exception of sterile processes are designed to meet a 4 sigma limit (which would permit as many as eight defects per 1000 units). The new approach is to center the process (in which the grand average is roughly equal to 100% of label potency or the target value of a given specification) and to reduce the process variability or noise around the mean or to achieve minimum variability by holding both to the new standard, batch after batch. In so doing, a 6 sigma limit may be possible (which is equivalent to not more than three to four defects per 1 million units), also called "zero defects." The goal of 6 sigma, "zero defects" is easier to achieve for liquid than for solid pharmaceutical dosage forms [16].
Process characterization represents the methods used to determine the critical unit operations or processing steps and their process variables, that usu¬ally affect the quality and consistency of the product outcomes or product attri¬butes. Process ranging represents studies that are used to identify critical process or test parameters and their respective control limits, which normally affect the quality and consistency of the product outcomes of their attributes. The follow¬ing process characterization techniques may be used to designate critical unit operations in a given manufacturing process.

A. Constraint Analysis
One procedure that makes subsystem evaluations and performance qualification trials manageable is the application of constraint analysis. Boundary limits of any technology and restrictions as to what constitutes acceptable output from unit operations or process steps should in most situations constrain the number of process variables and product attributes that require analysis.
A practical guide to constraint analysis comes to us from the application of the Pareto Principle (named after an Italian sociologist) and is also known as the 80-20 rule, which simply states that about 80% of the process output is governed by about 20% of the input variables and that our primary job is to find those key variables that drive the process.
The FDA in their proposed amendments to the CGMPs [17] have desig¬nated that the following unit operatioas are considered critical and therefore their processing variables must be controlled and not disregarded:
• Cleaning
• Weighing/measuring
• Mixing/blending
« Compression/encapsulation e Filling/packaging/labeling
B. Fractional Factorial Design
An experimental design is a series of statistically sufficient qualification trials that are planned in a specific arrangement and include all processing variables that can possibly affect the expected outcome of the process under investigation. In the case of a full factorial design, n equals the number of factors or process variables, each at two levels, i.e., the upper (+) and lower (-) control limits. Such a design is known as a In factorial. Using a large number of process variables (say, 9) we could, for example, have to run 29, or 512, qualification trials in order to complete the full factorial design.
The fractional factorial is designed to reduce the number of qualification trials to a more reasonable number, say, 10, while holding the number of ran¬domly assigned processing variables to a reasonable number as well, say, 9. The technique was developed as a nonparametric test for process evaluation by Box and Hunter [18] and reviewed by Hendrix [19]. Ten is a reasonable number of trials in terms of resource and time commitments and should be considered an upper limit in a practical testing program. This particular design as presented in Table 7 does not include interaction effects.

The guidelines on general principles of process validation [1] mention three options: (1) prospective process validation (also called premarket validation), (2) retrospective process validation, and (3) revalidation. In actuality there are four possible options.
A. Prospective Process Validation
In prospective process validation, an experimental plan called the validation protocol is executed (following completion of the qualification trials) before the process is put into commercial use. Most validation efforts require some degree of prospective experimentation to generate validation support data. This particu¬lar type of process validation is normally carried out in connection with the introduction of new drag products and their manufacturing processes. The for¬malized process validation program should never be undertaken unless and until the following operations and procedures have been completed satisfactorily:
1. The facilities and equipment in which the process validation is to
be conducted meet CGMP requirements (completion of installation
2. The operators and supervising personnel who will be "running" the
validation batch(es) have an understanding of the process and its re¬
3. The design, selection, and optimization of the formula have been
4. The qualification trials using (10 x size) pilot-laboratory batches have
been completed, in which the critical processing steps and process
variables have been identified, and the provisional operational control
limits for each critical test parameter have been provided
5. Detailed technical information on the product and the manufacturing
process have been provided, including documented evidence of prod¬
uct stability
6. Finally, at least one qualification trial of a pilot-production (100 x size)
batch has been made and shows, upon scale-up, that there were no
significant deviations from the expected performance of the process
The steps and sequence of events required to carry out a process validation assignment are outlined in Table 8. The objective of prospective validation is to prove or demonstrate that the process will work in accordance with a validation master plan or protocol prepared for pilot-product (100 x size) trials.
In practice, usually two or three pilot-production (100 x) batches are pre¬pared for validation purposes. The first batch to be included in the sequence

5. Failure to meet the requirements of the validation protocol with re¬spect to process inputs and output control should be subjected to re-qualification following a thorough analysis of process data and formal review by the CMC Coordination Committee.
B. Retrospective Validation
The retrospective validation option is chosen for established products whose manufacturing processes are considered stable and when on the basis of eco¬nomic considerations alone and resource limitations, prospective validation pro¬grams cannot be justified. Prior to undertaking retrospective validation, wherein the numerical in-process and/or end-product test data of historic production batches are subjected to statistical analysis, the equipment, facilities and subsys¬tems used in connection with the manufacturing process must be qualified in conformance with CGMP requirements. The basis for retrospective validation is stated in 21CFR 211.110(b): "Valid in-process specifications for such charac¬teristics shall be consistent with drug product final specifications and shall be derived from previous acceptable process average and process variability esti¬mates where possible and determined by the application of suitable statistical procedures where appropriate."
The concept of using accumulated final product as well as in-process nu¬merical test data and batch records to provide documented evidence of product/ process validation was originally advanced by Meyers [26] and Simms [27] of Eli Lilly and Company in 1980. The concept is also recognized in the FDA's Guidelines on General Principles of Process Validation [1].
Using either data-based computer systems [28,29] or manual methods, retrospective validation may be conducted in the following manner:
1. Gather the numerical data from the completed batch record and in¬
clude assay values, end-product test results, and in-process data.
2. Organize these data in a chronological sequence according to batch
manufacturing data, using a spreadsheet format.
3. Include data from at least the last 20-30 manufactured batches for
analysis. If the number of batches is less than 20, then include all
manufactured batches and commit to obtain the required number for
4. Trim the data by eliminating test results from noncritical processing
steps and delete all gratuitous numerical information.
5. Subject the resultant data to statistical analysis and evaluation.
6. Draw conclusions as to the state of control of the manufacturing pro¬
cess based on the analysis of retrospective validation data.
7. Issue a report of your findings (documented evidence).

One or more of the following output values (measured responses), which have been shown to be critical in terms of the specific manufacturing process being evaluated, are usually selected for statistical analysis.
1. Solid Dosage Forms
1. Individual assay results from content uniformity testing
2. Individual tablet hardness values
3. Individual tablet thickness values
4. Tablet or capsule weight variation
5. Individual tablet or capsule dissolution time (usually at t5m) or disinte¬
gration time
6. Individual tablet or capsule moisture content
2. Semisolid and Liquid Dosage Forms
1. pH value (aqueous system)
2. Viscosity
3. Density
4. Color or clarity values
5. Average particle size or distribution
6. Unit weight variation and/or potency values
The statistical methods that may be employed to analyze numerical output data from the manufacturing process are listed as follows:
1 Basic statistics (mean, standard deviation, and tolerance limits) [21]
2. Analysis of variance (ANOVA and related techniques) [21]
3. Regression analysis [22]
4. Cumulative sum analysis (CUSUM) [23]
5. Cumulative difference analysis [23]
6. Control charting (averages and range) [24,25]
Control charting, with the exception of basic statistical analysis, is proba¬bly the most useful statistical technique to analyze retrospective and concurrent process data. Control charting forms the basis of modern statistical process con¬trol.
C. Concurrent Validation
In-process monitoring of critical processing steps and end-product testing of current production can provide documented evidence to show that the manufac¬turing process is in a state of control. Such validation documentation can be provided from the test parameter and data sources disclosed in the section on retrospective validation.

Not all of the in-process tests enumerated above are required to demon¬strate mat the process is in a state of control. Selections of test parameters should be made on the basis of the critical processing variables to be evaluated.

The reader should realize that there is no one way to establish proof or evidence of process validation (i.e., a product and process in control). If the manufacturer is certain that its products and processes are under statistical con¬trol and in compliance with CGMP regulations, it should be a relatively simple matter to establish documented evidence of process validation through the use of prospective, concurrent, or retrospective pilot and/or product quality informa¬tion and data. The choice of procedures and methods to be used to establish validation documentation is left with the manufacturer.
This introduction was written to aid scientists and technicians in the phar¬maceutical and allied industries in the selection of procedures and approaches that may be employed to achieve a successful outcome with respect to product performance and process validation. The authors of the following chapters ex¬plore the same topics from their own perspectives and experience. It is hoped that the reader will gain much from the diversity and richness of these varied approaches.

D. Revalidation
Conditions requiring revalidation study and documentation are listed as follows:
1. Change in a critical component (usually refers to raw materials)
2. Change or replacement in a critical piece of modular (capital) equip¬
3. Change in a facility and/or plant (usually location or site)
4. Significant (usually order of magnitude) increase or decrease in batch
5. Sequential batches that fail to meet product and process specifications
In some situations performance requalification studies may be required prior to undertaking specific revalidation assignments.
The FDA process validation guidelines [1] refer to a quality assurance system in place that requires revalidation whenever there are changes in packag¬ing (assumed to be the primary container-closure system), formulation, equip¬ment or processes (meaning not clear) which could impact on product effective¬ness or product characteristics and whenever there are changes in product characteristics.
Approved packaging is normally selected after completing package perfor¬mance qualification testing as well as product compatibility and stability studies. Since in most cases (exceptions: transdermal delivery systems, diagnostic tests, and medical devices) packaging is not intimately involved in the manufacturing process of the product itself, it differs from other factors, such as raw materials.

1. Guidelines on General Principles of Process Validation, Division of Manufacturing
and Product Quality, CDER, FDA, Rockville, Maryland (May 1987).
2. Current Good Manufacturing Practices in Manufacture, Processing, Packing and
Holding of Human and Veterinary Drugs, Federal Register 43(190), 45085 and
45086, September 1978.
3. Good Manufacturing Practices for Pharmaceuticals, Willig, S. H. and Stoker, J.
R., Marcel Dekker, New York (1997).
4. Commentary, Pre-approval Inspections/Investigations, FDA, J. Parent. Sci. & Tech.
45:56-63 (1991).
5. Mead, W. J., Process validation in cosmetic manufacture, Drug Cosmet. bid., (Sep¬
tember 1981).
6. Chapman, K. G., A history of validation in the United States, Part I, Pharm. Tech.,
(November 1991).
7. Nash, R. A., The essentials of pharmaceutical validation in Pharmaceutical Dosage
Forms: Tablets, Vol. 3, 2nd ed., Lieberman, H. A., Lachman, L. and Schwartz, J.
В., eds., Marcel Dekker, New York (1990).
8. Nash, R. A., Product formulation, CHEMTECH, (April 1976).
9. Pharmaceutical Process Validation, Berry, I. R. and Nash, R. A., eds., Marcel
Dekker, New York (1993).

10. Nash, R. A., Making the Paper Match the Work, Pharmaceutical Formulation &
Quality (Oct/Nov 2000).
11. Guidance for Industry, Scale-Up & Postapproval Changes, CDER, FDA (Nov
12. Bala, G., An integrated approach to process validation, Pharm. Eng. 14(3) (1994).
13. Farkas, D. F., Unit operations optimization operations, CHEMTECH, July 1977.



14. Nash, R. A., Streamlining Process Validation, Amer. Pharm. Outsourcing May
15. Ishikawa, K., What is Total Quality Control? The Japanese Way, Prentice-Hall,
Englewood Cliffs, Ш (1985).
16. Nash, R. A., Practicality of Achieving Six Sigma or Zero-Defects in Pharmaceutical
Systems, Pharmaceutical Formulation & Quality, Oct./Nov. 2001.
17. CGMP: Amendment of Certain Requirements, FDA Federal Register, May 3,
18. Box, G. E. and Hunter, J. S., Statistics for Experimenters, John Wiley, N.Y. (1978).
19. Hendrix, C. D., What every technologist should know about experimental design,
CHEMTECH (March 1979).
20. Chapman, K. G., The PAR approach to process validation, Pharm. Tech., Dec.
21. Bolton, S., Pharmaceutical Statistics: Practical and Clinical Applications, 3rd ed.,
Marcel Dekker, New York (1997).
22. Schwartz, J. В., Optimization techniques in product formulation. J. Soc. Cosmet.
Chem. 32:287-301 (1981).
23. Butler, J. J., Statistical quality control, Chem. Eng. (Aug. 1983).
24. Deming, S. N.. Quality by Design, CHEMTECH, (Sept. 1988).
25. Contino, AV., Improved plant performance with statistical process control, Chem.
Eng. (July 1987).
26. Meyer, R. J., Validation of Products and Processes, PMA Seminar on Validation
of Solid Dosage Form Processes, Atlanta, GA, May 1980.
27. Simms, L., Validation of Existing Products by Statistical Evaluation, Atlanta, GA,
May 1980.
28. Agalloco, J. P., Practical considerations in retrospective validation, Pharm. Tech.
(June 1983).
29. Kahan, J. S., Validating computer systems, MD&DI (March 1987).

Process Validation
John M. Dietrick
C/.S. Food and Drug Administration, Rockvilie, Maryland, U.S.A.
Bernard T. Loftus
U.S. Food and Drug Administration, Washington, D.C., U.S.A.
Bernard T. Loftus was director of drug manufacturing in the Pood and Drug Administration (FDA) in the 1970s, when the concept of process validation was first applied to the phannaceutical industry and became an important part of current good manufacturing practices (CGMPs). His comments on the develop¬ment and implementation of these regulations and policies as presented in the first and second editions of this volume are summarized below [1].

The term process validation is not defined in the Food, Drug, and Cosmetic Act (FD&C) Act or in FDA's CGMP regulations. Many definitions have been of¬fered that in general express the same idea—that a process will do what it purports to do, or that the process works and the proof is documented. A June 1978 FDA compliance program on drug process inspections [2] contained the following definition:
This chapter was written by John M. Dietrick in his private capacity. No official support or endorse¬ment by the Food and Drug Administration is intended or should be inferred.

A validated manufacturing process is one which has been proved to do what it purports or is represented to do. The proof of validation is obtained through the collection and evaluation of data, preferably, beginning from the process development phase and continuing through the production phase. Validation necessarily includes process qualification (the qualifica¬tion of materials, equipment, systems, buildings, personnel), but it also in¬cludes the control on the entire process for repeated batches or runs.
The first drafts of the May 1987 Guideline on General Principles of Process Validation [3] contained a similar definition, which has frequently been used in FDA speeches since 1978, and is still used today: "A documented program which provides a high degree of assurance that a specific process will consistently pro¬duce a product meeting its pre-determined specifications and quality attributes."
Once the concept of being able to predict process performance to meet user requirements evolved, FDA regulatory officials established that there was a le¬gal basis for requiring process validation. The ultimate legal authority is Section 501(a)(2)(B) of the FD&C Act. [4], which states that a drug is deemed to be adulterated if the methods used in, or the facilities or controls used for, its manufacture, processing, packing, or holding do not conform to or were not operated or administrated in conformity with CGMP. Assurance must be given that the drug would meet the requirements of the act as to safety and would have the identity and strength and meet the quality and purity characteristics that it purported or was represented to possess. That section of the act sets the premise for process validation requirements for both finished Pharmaceuticals and active pharmaceutical ingredients, because active pharmaceutical ingredi¬ents are also deemed to be drugs under the act.
The CGMP regulations for finished pharmaceuticals, 21 CFR 210 and 211, were promulgated to enforce the requirements of the act. Although these regulations do not include a definition for process validation, the requirement is implicit in the language of 21 CFR 211.100 [5], which states: "There shall be written procedures for production and process control designed to assure that the drug products have the identity, strength, quality, and purity they purport or are represented to possess."
Although the emphasis on validation began in the late 1970s, the requirement has been around since at least the 1963 CGMP regulations for finished pharma¬ceuticals. The Kefauver-Harris Amendments to the FD&C Act were approved

in 1962 with Section 501(a)(2)(B) as an amendment. Prior to then, CGMP and process validation were not required by law. The FDA had the burden of prov¬ing that a drug was adulterated by collecting and analyzing samples. This was a significant regulatory burden and restricted the value of factory inspections of pharmaceutical manufacturers. It took injuries and deaths, mostly involving cross-contamination problems, to convince Congress and the FDA that a revi¬sion of the law was needed. The result was the Kefauver-Harris drug amend¬ments, which provided the additional powerful regulatory tool that FDA re¬quired to deem a drug product adulterated if the manufacturing process was not acceptable. The first CGMP regulations, based largely on the Pharmaceutical Manufacturers Association's manufacturing control guidelines, were then pub¬lished and became effective in 1963. This change allowed FDA to expect a preventative approach rather than a reactive approach to quality control. Section 505(d)(3) is also important in the implementation of process validation require¬ments because it gives the agency the authority to withhold approval of a new drug application if the "methods used in, and the facilities and controls used for, the manufacture, processing, and packing of such drug are inadequate to preserve its identity, strength, quality, and purity."
Another requirement of the same amendments was the requirement that FDA must inspect every drug manufacturing establishment at least once every 2 years [6]. At first, FDA did this with great diligence, but after the worst CGMP manufacturing situations had been dealt with and violations of the law became less obvious, FDA eased up its pharmaceutical plant inspection activi¬ties and turned its resources to more important problems.
The Drug Product Quality Assurance Program of the 1960s and 1970s involved first conducting a massive sampling and testing program of finished batches of particularly important drugs in terms of clinical significance and dollar volume, then taking legal action against violative batches and inspecting the manufacturers until they were proven to be in compliance. This approach was not entirely satisfactory because samples are not necessarily representative of all batches. Finished product testing for sterility, for example, does not assure that the lot is sterile. Several incidents refocused FDA's attention to process inspections. The investigation of complaints of clinical failures of several prod¬ucts (including digoxin, digitoxin, prednisolone, and prednisone) by FDA found significant content uniformity problems that were the result of poorly controlled manufacturing processes. Also, two large-volume parenteral manufacturers ex¬perienced complaints despite quality control programs and negative sterility test¬ing. Although the cause of the microbiological contamination was never proven, FDA inspections did find deficiencies in the manufacturing process and it be¬came evident that there was no real proof that the products were sterile.
What became evident in these cases was that FDA had not looked at the process itself—certainly not the entire process—in its regulatory activities; it was quality control- rather than quality assurance-oriented. The compliance officials were not thinking in terms of process validation. One of the first entries into process validation was a 1974 paper presented by Ted Byers, entitled "De¬sign for Quality" [7]. The term validation was not used, but the paper described an increased attention to adequacy of processes for the production of pharraa-ceuticals. Another paper—by Bernard Loftus before the Parenteral Drug Associ¬ation in 1978 entitled "Validation and Stability" [8]—discussed the legal basis for the requirement that processes be validated.
The May 1987 Guideline on General Principles of Process Validation [3] was written for the pharmaceutical, device, and veterinary medicine industries. It has been effective in standardizing the approach by the different parts of the agency and in communicating that approach to manufacturers in each industry.
As discussed in the preceding sections, process validation has been a legal re¬quirement since at least 1963. Implementation of the requirement was a slow and deliberate process, beginning with the development and dissemination of an agency policy by Loftus, Byers, and others, and leading to the May 1987 guide¬line. The guideline quickly became an important source of information to phar¬maceutical manufacturers interested in establishing a process validation pro¬gram. Many industry organizations and officials promoted the requirements as well as the benefits of validation. Many publications, such as Pharmaceutical Process Validation and various pharmaceutical industry journal articles, cited and often expanded on the principals in the guideline. During the same period, computer validation—or validation of computer controlled processes— also became a widely discussed topic in both seminars and industry publications. The regulatory implementation of the validation requirement was also a deliberate process by FDA. During the 1980s, FDA investigators often reported processes that had not been validated or had been inadequately validated. Batch failures were often associated with unvalidated manufacturing processes. The FDA issued a number of regulatory letters to deficient manufacturers citing the lack of adequate process validation as a deviation from CGMP regulations (21CFR 211.100), which causes the drug product to be adulterated within the meaning of Section 501(a)(2)(B) of the federal FD&C Act. Process validation was seldom the only deficiency listed in these regulatory letters. The failure of some manufacturers to respond to these early warnings resulted in FDA filing several injunction cases that included this charge in the early 1990s. Most of these cases resulted in consent decrees, and ultimately the adoption of satisfac¬tory process validation programs by the subject manufacturers. One injunction case filed in 1992, however, was contested in court and led to a lengthy written order and opinion by the U.S. District Court in February of 1993 [9]. The courtaffirmed the requirement for process validation in the current good manufactur¬ing regulations, and ordered the defendants to perform process validation studies on certain drug products, as well as equipment cleaning validation studies. This case and the court's ruling were widely circulated in the pharmaceutical industry and became the subject of numerous FDA and industry seminars.
The court also criticized the CGMP regulations for their lack of specific¬ity, along with their ambiguity and vagueness. Responding to this criticism, FDA drafted revisions to several parts of these regulations. The proposed revi¬sions were published in the Federal Register on May 3, 1996 [10]. One of the main proposed changes was intended to emphasize and clarify the process vali¬dation requirements. The proposal included a definition of process validation (the same definition used in the 1987 guideline), a specific requirement to vali¬date manufacturing processes, and minimum requirements for performing and documenting a validation study. These were all implied but not specific in the 1978 regulation. In proposing these changes, FDA stated that it was codifying current expectations and current industry practice and did not intend to add new validation requirements. Comments from all interested parties were requested under the agency's rule-making policies, and approximately 1500 comments were received. Most of the responses to the changes regarding process validation supported the agency's proposals, but there were many comments regarding the definitions and terminology proposed about which processes and steps in a pro¬cess should or should not require validation, the number of batches required for process validation, maintenance of validation records, and the assignment of responsibility for final approval of a validation study and change control deci¬sions. Because of other high-priority obligations, the agency has not yet com¬pleted the evaluation of these responses and has not been able to publish the final rule. In addition to the official comments, the proposed changes prompted numerous industry and FDA seminars on the subject.
Process validation is not just an FDA or a U.S. requirement. Similar re¬quirements are included in the World Health Organization (WHO), the Pharma¬ceutical Inspection Co-operation Scheme (PIC/S), and the European Union (EU) requirements, along with those of Australia, Canada, Japan, and other interna¬tional authorities.
Most pharmaceutical manufacturers now put substantial resources into process validation for both regulatory and economic reasons, but despite contin¬ued educational efforts by both the agency and the pharmaceutical industry, FDA inspections (both domestically and internationally) continue to find some firms manufacturing drug products using unvalidated or inadequately validated processes. Evidently there is still room for improvement, and continued discus¬sion, education, and occasional regulatory action appears warranted.
The future of process validation is also of great interest, especially with the worldwide expansion of pharmaceutical manufacturing and the desire for

6 Dietrick and Loftus
harmonized international standards and requirements. Many manufacturers are also working on strategies to reduce the cost of process validation and incorpo¬rate validation consideration during product design and development. New tech¬nologies under development for 100% analysis of drug products and other inno¬vations in the pharmaceutical industry may also have a significant effect on process validation concepts and how they can be implemented and regulated.

1. Loftus, В. Т., Nash, R. A., ed. Pharmaceutical Process Validation, vol. 57. New
York: Marcel Dekker (1993).
2. U.S. Food and Drag Administration. Compliance Program no. 7356.002.
3. U.S. Food and Drug Administration. Guideline on General Principles of Process
Validation. Rockville, MD: FDA, 1987.
4. Federal Food Drug and Cosmetic Act, Title 21 U.S. Code, Section 501 (a)(2)(B).
5. Code of Federal Regulations, Title 21, Parts 210 & 211. Fed Reg 43, 1978.
6. U.S. Code, Federal Food Drug and Cosmetic Act, Title 21, Section 510 (h).
7. Byers, Т. Е. Design for quality, Manufacturing Controls Seminar, Proprietary Asso¬
ciation, Cherry Hill, NJ, Oct. 11, 1974.
8. Loftus, В. Т. Validation and stability, meeting of Parenteral Drug Association,
9. U.S. v. Barr Laboratories, Inc., et al., Civil Action No. 92-1744, U.S. District Court
for the District of New Jersey, 1973.
10. Code of Federal Regulations, Title 21, Parts 21 & 211, Proposed Revisions, Fed Reg (May 3, 1996).

Aden Y. Chao
Watson Labs, Carona, California, U.S.A.
F, St. John Forbes
Wyeth Labs, Peart River, New York, U.S.A.
Reginald F. Johnson and Paul Von Doehren
Searie & Co., Inc., Skokie, Illinois, U.S.A.
Validation is an essential procedure that demonstrates that a manufacturing pro¬cess operating under defined standard conditions is capable of consistently pro¬ducing a product that meets the established product specifications. In its proposed guidelines, the U.S. Food and Drug Administration (FDA) has offered the following definition for process validation [1].
Process validation is establishing documented evidence that provides a high degree of assurance that a specific process (such as the manufacture of pharmaceutical dosage forms) will consistently produce a product meeting its predetermined specifications and quality characteristics.
Many individuals tend to think of validation as a stand-alone item or an afterthought at the end of the entire product/process development sequence. Some believe that the process can be considered validated if the first two or three batches of product satisfy specifications.
Prospective validation is a requirement (Part 211), and therefore it makes validation an integral part of a carefully planned, logical product/process devel¬opmental program. An outline of the development sequence and requirements relevant to process validation is presented in Figure 1. After briefly discussing organizational aspects and documentation, the integration of validation into the product development sequence is discussed. At the end of the chapter there is a

8 Chao et al.
brief discussion of specific ways in which experimental programs can be defined to ensure that critical process development and validation objectives are met.
Prospective validation requires a planned program and organization to carry it to successful completion. The organization must have clearly defined areas of responsibility and authority for each of the groups involved in the program so that the objective of validating the process can be met. The structure must be tailored to meet the requirements in the specific organization, and these will vary from company to company. The important point is that a defined structure exists, is accepted, and is in operation. An effective project management struc¬ture will have to be established in order to plan, execute, and control the pro¬gram. Without clearly defined responsibilities and authority, the outcome of process validation efforts may not be adequate and may not comply with CGMP requirements.
An effective prospective validation program must be supported by documenta¬tion extending from product initiation to full-scale production. The complete documentation package can be referred to as the master documentation file.
It will accumulate as a product concept progresses to the point of being placed in full-scale production, providing as complete a product history as possi¬ble. The final package will be the work of many individual groups within the organization. It will consist of reports, procedures, protocols, specifications, ana¬lytical methods, and any other critical documents pertaining to the formulation, process, and analytical method development. The package may contain the ac¬tual reports, or it may utilize cross-references to formal documentation, both internal and external to the organization.
The ideal documentation package will contain a complete history of the final product that is being manufactured. In retrospect, it would be possible to trace the justification or rationale behind all aspects of the final product, process, and testing.
The complete master documentation file not only provides appropriate rationale for the product, process, and testing, but also becomes the reference source for all questions relating to the manufacture of a product at any plant location. This master documentation file, however, should not be confused with the concept of the master product document, which is essential for routine manu¬facturing of the product and is described later in the chapter. The master documentation file should contain all information that was generated during the en¬tire product development sequence to a validation process.
Product development usually begins when an active chemical entity has been shown to possess the necessary attributes for a commercial product. The product development activities for the active chemical entity, formulation, and process form the foundation upon which the subsequent validation data are built.
Generally, product development activities can be subdivided into formula¬tion and process development, along with scale-up development.
A. Formulation Development
Formulation development provides the basic information on the active chemical, the formula, and the impact of raw materials or excipients on the product. Typi¬cal supportive data generated during these activities may include the following:
1. Preformulation profile or characterization of the components of the formula, which includes all the basic physical or chemical information about the active pharmaceutical ingredients (API, or the chemical entity) and excipients
2. Formulation profile, which consists of physical and chemical characteristics required for the products, drugexcipient compatibility studies, and the effect of formulation on in vitro dissolution
3. Effect of formulation variables on the bioavailability of the product
4. Specific test methods
5. Key product attributes and/or specifications
6. Optimum formulation
7. Development of cleaning procedures and test methods
Formulation development should not be considered complete until all those fac¬tors that could significantly alter the formulation have been studied. Subsequent minor changes to the formulation, however, may be acceptable, provided they are thoroughly tested and are shown to have no adverse effect on product.
B. Process Development
Even though the process development activities typically begin after the formu¬lation has been developed, they may also occur simultaneously. The majority of the process development activities occur either in the pilot plant or in the proposed manufacturing plant. The process development program should meet the following objectives:
1. Develop a suitable process to produce a product that meets all
a. Product specifications
b. Economic constraints
с Current good manufacturing practices (CGMPs)
2. Identify the key process parameters that affect the product attributes
3. Identify in-process specifications and test methods
4. Identify generic and/or specific equipment that may be required
It is important to remember that cleaning procedures should at least be in the final stages of development, as equipment and facilities in the pilot or proposed manufacturing plant are involved, and the development of the cleaning verifica¬tion test methods must be complete.
Process development can be divided into several stages.
Challenging of critical process parameters
Verification of the developed process

1. Design
This is the initial planning stage of process development. The design of the process should start during or at the end of the formulation development to define the process to a large extent. One aspect of the process development to remember is end user (manufacturing site) capabilities. In other words, the practicality and the reality of the manufacturing operation should be kept in perspective. Process must be developed in such a manner that it can easily be transferred to the manufacturing site with minimal issues. During this stage, technical operations in both the manufacturing and quality control departments should be consulted.
Key documents for the technical definition of the process are the flow diagram, the cause-and-effect diagram, and the influence matrix. The details of the cause-and-effect diagram and the influence matrix will be discussed under experimental approach in a later section.
The flow diagram identifies all the unit operations, the equipment used, and the stages at which the various raw materials are added. The flow diagram in Figure 3 outlines the sequence of process steps and specific equipment to be used during development for a typical granulation solid dosage from product. The flow diagram provides a convenient basis on which to develop a detailed list of variables and responses.

Preliminary working documents are critical, but they should never be cast in stone, since new experimental data may drastically alter them. The final ver¬sion will eventually be an essential part of the process characterization and technical transfer documents.
Regardless of the stage of formulation/process development being consid¬ered, a detailed identification of variables and responses is necessary for early program planning. Typical variables and responses that could be expected in a granulated solid dosage form are listed in Table 1. This list is by no means complete and is intended only as an example.

As the developmental program progresses, new discoveries will provide an update of the variables and responses. It is important that current knowledge be adequately summarized for the particular process being considered. It should be pointed out, however, that common sense and experience must be used in evaluating the variables during process design and development. An early trans¬fer of the preliminary documentation to the manufacturing and quality control departments is essential, so that they can begin to prepare for any new equip¬ment or facilities that may be required.
2. Challenging of Process Parameters
Challenging of process parameters (also called process ranging) will test whether or not all of the identified process parameters are critical to the product and process being developed. These studies determine:
The feasibility of the designed process The criticality of the parameters
This is usually a transition stage between the laboratory and the projected final process. Figure 4 also shows typical responses that may have to be evalu¬ated during the ranging studies on the tableted product.
3. Challenging of Critical Process Parameters'or
Characterization of the Process
Process characterization provides a systematic examination of critical variables found during process ranging. The objectives of these studies are
Confirm critical process parameters and determine their effects on product
quality attributes.
Establish process conditions for each unit operation. Determine in-process operating limits to guarantee acceptable finished
product and yield. Confirm the validity of the test methods.
A carefully planned and coordinated experimental program is essential in order to achieve each of these objectives. Techniques to assist in defining experimental programs are mentioned later in the chapter.
The information summarized in the process characterization report pro¬vides a basis for defining the full-scale process.
4. Verification
Verification is required before a process is scaled up and transferred to produc¬tion. The timing of this verification may be critical from a regulatory point of view, as the there is little or no room for modifying the parameter values and specifications, particularly shifting or expanding after the regulatory submission is made. This ensures that it behaves as designed under simulated production conditions and determines its reproducibility. Key elements of the process verifi¬cation runs should be evaluated using a well-designed in-process sampling pro¬cedure. These should be focused on potentially critical unit operations. Vali¬dated in-process and final-product analytical procedures should always be used. Sufficient replicate batches should be produced to determine between- and within-batch variations.
Testing during these verification runs will be more frequent and cover more variables than would be typical during routine production. Typically the testing requirements at the verification stage should be the same or more than the proposed testing for process validation runs.

18 Chaoetal.
A. Scale-Up Studies
The transition from a successful pilot-scale process or research scale to a full-scale process requires careful planning and implementation. Although a large amount of information has been gathered during the development of the process (i.e., process characterization and process verification studies), it does not neces¬sarily follow that the full-scale process can be completely predicted.
Many scale-up parameters are nonlinear. In fact, scale-up factors can be quite complex and difficult to predict, based only on experience with smaller-scale equipment. In general, the more complex the process, the more complex the scale-up effect.
For some processes, the transition from pilot scale or research scale to full scale is relatively easy and orderly. For others the transition is less predictable. More often than not there will be no serious surprises, but this cannot be guaran¬teed. Individuals conducting the transfer into production should be thoroughly qualified on both small- and large-scale equipment.
The planning for scale-up should follow the same general outline followed for process characterization and verification. It usually begins when process development studies in the laboratory have successfully shown that a product can be produced within specification limits for defined ranges of process param¬eters.
Frequently, because of economic constraints, a carefully selected excipient may be used as a substitute for the expensive active chemical in conducting initial scale-up studies. Eventually, the active chemical will have to be used to complete the scale-up studies, however.
It is common sense that every effort will be made to conduct the final scale-up studies under CGMP conditions, thus any product produced with speci¬fications can be considered for release as a finished salable product (for over-the-counter products only).
B. Qualification Trials
Once the scale-up studies have been completed, it may be necessary to manufac¬ture one or more batches at full scale to confirm that the entire manufacturing process, comprising several different unit operations, can be carried out smoothly. This may occur prior to or after the regulatory submission, depending on the strategy used in filing.
C. Process Vaiidation Runs
After the qualification trials have been completed, the protocol for the full-scale process validation runs can be written. Current industry standard for the valida¬tion batches is to attempt to manufacture them at target values for both process

D. Master Product Document
An extensive quantity of documents is generated at each stage of the develop¬ment arid validation of the final production process. Some of these documents will be directly related to the manufacture of the final products. Others may provide the basis for decisions that ultimately result in the final process.
The documents that are required for manufacturing the product then be¬come the master product document. This document must be capable of provid¬ing all of the information necessary to set up the process to produce a product consistently and one that meets specifications in any location.

The objective in this section is to examine experiments or combinations of re¬lated experiments that make up development programs so that adequate justifi¬cation can be developed for the formulation, process, and specifications. The emphasis will be on techniques to increase developmental program effective¬ness.
A logical and systematic approach to each experimental situation is essen¬tial. Any experiment that is performed without first defining a logical approach is certain to waste resources. The right balance between overplanning and under-planning should always be sought.
It is usually impossible to define a substantial experimental effort at the beginning and then execute it in every detail without modification. To overcome this, it is convenient to split the program into a number of stages.
Each stage will normally consist of several specific experiments. The ear¬lier experiments tend to supply initial data concerning the process and define preliminary operating ranges for important variables. As results become avail¬able from each stage, they can be used to assist in defining subsequent stages in the experimental program. In some cases it may be necessary to redefine completely the remainder of the experimental program on the basis of earlier results.
The following discussion describes some techniques to help improve ex¬perimental program effectiveness. A logical and systematic approach coupled with effective communication among individuals associated with the program is emphasized. Topics to be discussed include
Defining program scope Process summary Experimental design and analysis Experiment documentation Program organization
A. Program Scope
Defining a clear and detailed set of objectives is a necessary first step in any experimental program. Some similarity exists between objectives for different products and processes using similar existing technology. For products and pro¬cesses at the forefront of technology, the definition of specific experimental objectives can be a continuing activity throughout product development.
Constraints on planning experimental programs can be classified accord¬ing to their impact on time, resources, and budget. The effect and impact of these should be incorporated into the experimental program early to avoid com¬promising critical program objectives.

B. Process Summary
An initial clear understanding of the formulation and/or process is important. The following techniques can assist in summarizing current process knowledge.
1. Flow Diagram
A process flow diagram (Fig. 3) can often provide a focal point of early program planning activities. This diagram outlines the sequence of process steps and specific equipment to be used during development for a typical granulated prod¬uct. Flow diagram complexity will depend on the particular product and process. The flow diagram provides a convenient basis on which to develop a detailed list of variables and responses.
2. Variables and Responses
For process using existing technology, many of the potential variables and re¬sponses may have already been identified in previous product-development studies or in the pharmaceutical literature. Once properly identified, the list of variables and responses for the process is not likely to change appreciably. Typi¬cal variables and responses that could be expected in a granulated solid dosage form are listed in Table 1.
In addition, the relative importance of variables and responses already identified will likely shift during development activities.
3. Cause-ancl-Effect Diagram
An efficient representation of complex relationships between many process and formulation variables (causes), and a single response (effect) can be shown by using a cause-and-effect diagram [1]. Figure 4 is a simple example.
A central arrow in Figure 4 points to a particular single effect. Branches off the central arrow lead to boxes representing specific process steps. Next, principle factors of each process step that can cause or influence the effect are drawn as subbranches of each branch, until a complete cause-and-effect diagram is developed. This should be as detailed a summary as possible. An example of a more complex cause-and-effect diagram is illustrated in Figure 5. A separate summary for each critical product characteristic (e.g., weight variation, dissolu¬tion, friability) should be made.
4. Influence Matrix
Once the variables and responses have been identified, it is useful to summarize their relationships in an influence matrix format, as shown in Figure 6. Based on the available knowledge, each process variable is evaluated for its potential
effects on each of the process responses or product characteristics. The strength of the relationship between variables and responses can be indicated by some appropriate notation, such as strong (S), moderate (M), weak (W), or none (N), together with special classifications such as unknown (?).
Construction of the influence matrix assists in identifying those variables with the greatest influence on key process or product characteristics. These vari¬ables are potentially the most critical for maintaining process control and should be included in the earliest experiments. Some may continue to be investigated during development and scale-up.
Many different experimental designs and analysis methods can be used in devel¬opment activities (Fig. 7). Indeed, the possibilities could fill several books. For¬tunately, in any given situation, it is not necessary to search for that single design or analysis method that absolutely must be used; there are usually many possibilities. In general, designs that are usable offer different levels of effi¬ciency, complexity, and effectiveness in achieving experimental objectives.
It is not possible to list specific designs that will always be appropriate for general occasions. Any attempt to do so would be sure to be ineffective, and the uniqueness of individual experimental situation carefully, including
Specific objectives
Available resources
Availability of previous theoretical results
Relevant variables and responses
Qualifications and experience of research team members
Cost of experimentation
It should also be determined which design is appropriate. A statistician who is experienced in development applications can assist in suggesting and evaluating candidate designs. In some cases, the statistician should be a full-time member
of the research team.

C. Experiment Documentation
Documentation is essential to program planning and coordination, in addition to the obvious use for the summary of activities and results. Written communica¬tion becomes important for larger complex programs, especially when con¬ducted under severe constraints on time and resources. Documentation can con¬sist of some or all of the following items:
1. Objectives; an exact statement of quantifiable results expected from the experiment
2. Experimental design; a detailed list of the experimental conditions to be studied and the order of investigation
3. Proposed/alternate test methods
a. A list of test methods consistent with the type of experiment being performed
b. A detailed description of the steps necessary to obtain a valid measurement
c. Documentation supporting the accuracy, precision, sensitivity, and so on of the test methods
4. Equipment procedures; documentation of safety precautions and step by-step methods for equipment setup, operation, and cleanup
5. Sampling plans; the type, number, location, and purpose of samples
to be taken during the experiment; in addition, the type and number
of all measurements to be performed on each sample

6. Protocol; a formal written experimental plan that presents the aforementioned experimental documentation in a manner suitable for review
7. Data records
a. Experiment log; details of events in the experiment noting process adjustments and any unusual occurrences
b. In-process measurements; records of the magnitude of critical process parameters during the experimental sequence
Sample measurements; recorded values of particular measurements on each sample
8. Report; documentation of experiment implementation, exceptions/modifications to the protocol, results, and conclusion

D. Program Organization
Throughout the experimental phases of the development program, it is essential to maintain effective communication among various team members. This is fa¬cilitated by having one individual with the necessary technical and managerial skills assume responsibility for the experimental program, including procuring resources and informing management of progress.
In a large experimental program, the responsible individual may serve as a project leader or manager with little or no technical involvement.
Prospective validation of a production process utilizes information generated during the entire development sequence that produced the final process.
Validation is supported by all phases of development from the product concept.
As a potential product moves through the various developmental stages, information is continually generated and incorporated into a master documentation file. When the validation runs are planned for the final process, they will be based on the master documentation file contents. The information generated during the validation runs is usually the last major item to go into the master documentation file.
An abstract of the master documentation file is the master product docu¬ment, which is the source of all information required to set up the process at any location.
Though validation may seem to be a stand-alone item, it actually is an integral portion of the entire product/process development sequence.
1. FDA. Guidelines on General Principles of Process Validation. Rockville, MD: Division of Manufacturing and Product Quality (HFN-320) Center for Drags and Biologies (May 1987).
2. Box, G. E. P., Gunter, W. G., and Hunter, J. S. Statistics for Experimenters: An Introduction to Design, Data Analysis, and Model Building. New York: Wiley (1978).
3. Box, G. E. P., and Draper, N. R. Evolutionary Operation: A Statistical Method for Process Improvement, New York: Wiley (1969).
4. Cornell, J. A. Experiments with Mixtures: Design, Models and the Analysis of Mixture Data. New York: Wiley (1981).

A. Product Selection Criteria for Retrospective Validation
For a product to be considered for retrospective validation, it must have a stable process; that is, one in which the method of manufacture has remained essen¬tially unchanged for a period of time.
The first step in the product selection process is therefore to obtain a summary of changes in the method of manufacture. In most companies such information is part of the master batch record file. Then a time interval is se¬lected that represents the last 20 to 30 batches. Products for which there is no record of a change in the method of manufacture or control during this period can be regarded as candidates for validation. The 20-to-30-batch rule originates from control chart principals, which consider 20 to 30 points that plot within the limits as evidence of a stable process [4]. Once this criterion is met, the number selected is actually somewhat arbitrary, as there is no one number that is correct for every product. The ideal number of batches required to study a product is theoretically the number that permits all process variables to come into play. By process variables, we mean raw materials from different but ap¬proved vendors, introduction of similar but different pieces of equipment, per¬sonnel and seasonal changes, and the like. This academic approach may present a rather unwieldy situation, especially for a high-volume product, for which change in process variables occurs infrequently. The influence of seasonal changes is such an example. In such instances, compromise will need to be reached between process variables included for study and the number of batches that can be examined for data. This decision making is best handled by a valida¬tion committee, the organization and makeup of which is covered in detail later in this chapter.
The second step in the product selection process addresses the situation in which a change in the method of manufacture or control was implemented dur¬ing the last 20 or so production batches. The fact that a change has occurred does not automatically disqualify the product for retrospective validation. One must first know whether the particular modification has caused an expected result to be different to the extent that it is no longer comparable to previous batches. An example may be helpful. Suppose the method of granulating was changed midway through the series of 20 batches selected for the validation study. The number of batches representing the new process would be signifi¬cantly reduced and could be insufficient to capture some of the interactions that can affect process reproducibility. In general, a history of any one of the following changes to the method of manufacture and control should be fully investi¬gated before any decision is made to validate retrospectively:
1. Formulation changes involving one or more of the active ingredients or key excipients
2. Introduction of new equipment not equivalent in every respect to that previously in use
3. Changes in the method of manufacture that may affect the product's characteristics
4. Changes to the manufacturing facility
A product found to be unsuitable for retrospective validation because of a revised manufacturing process is a likely candidate for prospective validation, which is beyond the scope of this chapter [1]. Such a discovery, however, should be brought to the attention of the appropriate authority. In today's regulatory environment ignoring the matter would be imprudent.
The third and last step in our selection process is to identify which prod¬ucts are likely to be discontinued because of a lack of marketing interest or regulatory consideration, to be sold, or to be reformulated. The timing of these events will dictate whether the product in question remains a viable candidate for retrospective validation.
The foremost discussion on developing a list of suitable products for study is summarized in Figure 1.
B. Organizing for Retrospective Validation
To this point we have produced a list of products that may be validated retro¬spectively; that is, their manufacturing processes are relatively stable, and so adequate historical data exist on which to base an opinion. The next consider¬ation is the formal mechanism for validating the individual products. Appro¬priate organizational structures for effectively validating processes have been put forth, but mostly in conjunction with the validation of new product introduc¬tions. Still, these recommendations can serve as models. Because the products being studied are marketed products, the quality assurance and production de¬partments can be expected to make major contributions. In fact, as far as retro¬spective validation is concerned, it may be more appropriate for one of these departments to coordinate the project. The research and engineering depart¬ments, of course, will be needed, especially where recent process changes have been encountered or equipment design is at issue.
The time requirements dictate that such work be assigned to a function with discretionary time, possibly a technical services group or a quality engineer. Management commitment is especially crucial if disruptive influences are to be minimized. The loss of a committee member to another project is such an example.
The various activities and responsibilities associated with retrospectively vali¬dating a product must be put in writing. All too often this simple but crucial step is omitted for the sake of expediency only to find at a later date that the initial assumptions cannot be recalled. Aside from maintaining consistency, a written procedure to describe the work being performed satisfies the intent of the current good manufacturing practice (CGMP) regulations.
In general, the written operating procedure should delineate in reasonable detail how the validation organization will function. Not every situation can be anticipated, and this should not be the goal. There should be sufficient detail, however, to ensure consistency of performance in an undertaking that may con¬tinue for several months. In the preparation of such a document, the following questions should be answered:
1. Which organizational functions will be represented on the validation committee?
2. What mechanism exists for validation protocol preparation and approval?
3. What criteria are used to select critical process steps and quality control tests for which data will be collected?
4. How often will the committee meet to ensure prompt evaluation of study data?
5. Who has responsibility for documenting committee decisions? For report preparation?
6. Is there a provision for follow-up in the event of unexpected findings?
7. Where will the original study data and reports be archived?
In the preceding discussion of areas of interest to the validation organiza¬tion, two concepts were introduced that deserve further clarification: (1) critical process steps and quality control tests that characterize the operation, and (2) validation protocol.
1. Critical Process Steps and Control Tests
Critical process steps are operations performed during dosage-form manufacture that can contribute to variability of the end product if not controlled. Since each type of dosage form requires different machinery and unit operations to produce the end product, the critical process steps will also differ. For each product considered suitable for retrospective validation, a list of these steps must be compiled following careful analysis of the process by technically competent persons. In a similar manner, in-process and finished-product tests should be screened to identify those that may be of some value. As a rule, tests in that the outcome is quantitative will be of greatest interest.
A flow diagram of the entire operation, but particularly of the manufactur¬ing process, may be helpful in identifying critical steps, especially where the process involves many steps. Such a diagram is also a useful addition to the validation report prepared at the conclusion of the study.

2. Validation Protocol
A written protocol that describes what is to be accomplished should be prepared [5]. It should specify the data to be collected, the number of batches to be included in the study, and how the data, once assembled, will be treated for relevance. The criteria for acceptable results should be described. The date of approval of the protocol by the validation organization should also be noted. The value of a protocol is to control the direction of the study, as well as provide a baseline in the event unanticipated developments necessitate a change in strat¬egy. A written protocol is also an FDA recommendation .
D. Other Considerations
Comprehensive records of complaints received either directly from the customer or through a drug problem reporting program should be reviewed. Furthermore, a record of any follow-up investigation of such complaints is mandatory [6] and should be part of this file. Review of customer complaint records can furnish a useful overview of process performance and possibly hint at product problems. Complaint analysis should therefore be viewed as a meaningful adjunct to the critical process step and control test selection process.
Batch yield reflects efficiency of the operation. Because yield figures are the sum of numerous interactions, they fail in most cases to provide specific information about process performance and therefore must be used with caution in retrospective validation. In any event, this information should be collected, as it can contribute to further refinement of the yield limits that appear in the batch record.
Lot-to-lot differences in the purity of the therapeutic agent must be consid¬ered when evaluating in-process and finished-product test results. In addition to potency such qualities as particle size distribution, bulk density, and source of the material will be of interest. Such information should be available from the raw material test reports prepared by the quality control laboratory for each lot of material received. The physical characteristics of the excipients should not be overlooked, especially for those materials with inherent variability. Metallic stearates is a classic example. In such instances, the source of supply is desirable
information to have available.
There is value in examining logs of equipment and physical plant mainte¬nance. These documents can provide a chronological profile of the operating environment and reveal recent alterations to the process equipment that may have enough impact to disqualify the product from retrospective validation con¬sideration. For this reason, it is always prudent to contemplate equipment status early in the information-gathering stage. The availability of such information should be ascertained for yet another reason: rarely is equipment dedicated to one product. More often than not, each blender, comminutor, tablet press, and so forth is used for several operations. Information gathered initially can there¬fore be incorporated into subsequent studies.
Retrospective validation is directed primarily toward examining the rec¬ords of past performance, but what if one of these documents is not a true reflection qf the operation performed? Suppose that changes have crept into the processing operation over time and have gone unreported. This condition would result in the validation of a process that in reality does not exist. It is therefore essential to audit the existing operation against the written instructions. There is obvious advantage to undertaking this audit before commencing data acquisi¬tion. Ideally, the manufacture of more than one batch should be witnessed, espe¬cially where multiple-shift operations are involved. The same logic would apply to the testing performed in process and at the finished stage. If any deviation from the written directions is noted, an effort must be made to measure its impact. In this regard, the previously described validation organization is a logi¬cal forum for discussion and evaluation.
As a rule, batches that are rejected or reworked are not suitable for inclu¬sion in a retrospective validation study [7]. Indeed, a processing failure that is not fully explainable should be cause to rethink the application of retrospective validation. Nonconformance to specification that is attributable to a unique event-operator error, for example, may be justifiably disregarded. In such cases, the batch is not considered when the historical data are assembled.
Raw materials, both actives and excipients, can be a source of product variability. To limit this risk, there should be meaningful acceptance specifica¬tions and periodic confirmation of test results reported on the supplier's certifi¬cate of analysis. Also, purchases must be limited to previously qualified suppli¬ers. A determination that such controls are in place should be part of any retrospective validation effort.
The following discussion will focus on how to apply the previously discussed concepts to the validation of marketed products. To provide a fuller understand¬ing of this procedure, the manufacture of several dosage forms designed for different routes of administration will be examined. For each dosage form, criti¬cal process steps and quality control tests will be identified. Useful statistical techniques for examining the assembled data will be illustrated. It is also impor¬tant to note that not all of the collected information for a product lends itself to this type of analysis. This will become more apparent as we proceed with the evaluation of the five drugs under consideration.

The second major step is granulation. The process is controlled by the operator, whose judgment is relied on for the appropriate end point. As no information useful for process validation is available, we will move on to the next step, comminution.
The batch record calls for passing the wet mix through a comminutor using a no. 5 or 7 drilled stainless steel screen. Knife position and rotational speed are two other factors that influence particle size; however, the step instruc¬tion is quite specific about machine setup. Therefore, only screen size is a source of variability for this step. We will want to know the frequency of use of each screen.
Next, the granulation is dried to a target moisture of 1%. Either a tray or fluid bed dryer may be used, at the discretion of area supervision. Regardless of the method, drying time will be of interest. In addition, the final moisture content should be ascertained for each batch. The dried granulation and lubricant are then oscillated using a no. 10 or 12 wire screen. This is the last sizing operation of the process; it will determine the particle size distribution of the final blend. Knowing the history of use of each screen size is thus important.
The lubricant and granulation are blended for several minutes. The elapsed mixing time is of interest because of its impact on drug distribution and the generally deleterious effect of the lubricant on dissolution.
Because excess moisture is thought to have a negative effect on the dosage form, loss on drying (LOD) is determined on the final blend.
Blending is followed by tableting. During compression, online measure¬ments such as tablet weight, hardnessJ and disintegration are made by the pro¬cess operator in order to ensure uniformity of the tablets. The weight of the tablets is not measured individually; rather, the average weight of 10 tablets is recorded. Although these data are good indicators of operation and machine performance, we would prefer to have the more precise picture provided by individual tablet weight.
Disintegration time and tablet hardness data could be collected from the manufacturing batch records; however, for ease of administration these figures will be obtained from the quality control test results, which also contain individ¬ual tablet weighings.
Disintegration time was selected as a critical variable because for a drug substance to be absorbed it must first disintegrate and then dissolve. The resis¬tance of a tablet to breakage, chipping, and so forth depends on its hardness.

Disintegration, too, can be influenced by hardness of the tablet. For these rea¬sons, hardness testing results also will be examined.
Specifications used by quality control to release drug A are found in a laboratory procedure. In addition to the previously discussed hardness and disin¬tegration time requirements, the procedure calls for determining the average tablet weight by the United States Pharmacopeia (USP) procedure; that is, 20 individual tablets are weighed.
The control procedure also requires assay of individual tablets. Of all the information available, these data will be the most useful in reaching an opinion of the adequacy of the process to distribute the therapeutic agent uniformly.
In addition, the laboratory checks the moisture content of the bulk tablets. It will be interesting to compare these results to me LOD of the final blend to measure the contribution of material handling.
Critical manufacturing steps and quality control tests for drug A, identified as a result of the review, are summarized in Table 1.
1. Evaluation of Historical Data
Earlier in the discussion of process validation strategies, 20 production batches were suggested as a minimum number upon which to draw conclusions about the validity of the process. In this particular example, however, two distinct methods of drying are provided. In order to have sufficient history on each operation, the number of batches examined was increased to 30.
The batches were selected so that the same number was dried by each process. For the other critical manufacturing steps and release tests listed in Table 1, data were collected for all 30 batches.
The first manufacturing step, premix blending time, was consistently re¬ported as 10 min, but with one exception. In this instance, the powders were tumbled for 20 min, which is still within the limits (10 to 20 min) prescribed by the batch record. It would be interesting to know if this source of variability can materially affect attributes of the final product. Unfortunately, having only one batch produced by the 20-min process does not permit statistically valid comparisons. At best, test results for the single 20-min batch can be screened using summary data from the remainder of the study. Under different circum¬stances, batches would have been grouped by mixing time and compared by dosage form attributes. More than likely, subsequent manipulation of the blend would have negated any contribution, allowing us to conclude that a mixing time of 10 to 20 min is not unreasonable.
At the wet milling step we encounter a situation similar to preblending; that is, only two of the 30 study batches are prepared using the no. 5 drilled screen. The no. 7 is obviously the screen of choice. The purpose of this step is to produce particles of reasonably uniform size, which in turn will improve drying. From the records, we also know that the no. 5 screen was used only with batches that were tray dried. Elapsed drying time and residual moisture were compared for the two batches from the no. 5 screen process and the other 13 batches that were tray dried. No important differences were detected. Still, in light of the limited use of the no. 5 screen, it would not be inappropriate to recommend this option be eliminated from the processing instructions.
Mean drying time for the oven tray process is 19.2 hr. All 15 batches were dried within the specified time of 16 to 20 hr. No seasonal influence was apparent. The average moisture content of these batches is 1.2%; the standard deviation is 0.3%. The 15 batches dried using the fluid bed dryer had a residual moisture of 0.8% (SD = 0.1%). Drying time is mechanically controlled and not recorded. The statistics favor the fluid bed process; it is more efficient and uniform. There is nothing in these data to disqualify the oven tray dryer from further use, however.
Oscillation of the dried granulation and lubricant was accomplished in every instance using a no. 10 wire screen. Reference to the no. 12 screen, the alternative method for pulverizing the batch, must be deleted from the manufac¬turing instructions for the process to be validated retrospectively.
The final mix blending time was reported as either 10 or 15 min. Twenty-one of the 30 batches were tumbled for 10 min and the remainder were mixed for 15 min. The mixing time is not mechanically controlled or automatically recorded; it is left to the operator to interpret elapsed time. Because of the importance of the step to distribution of the therapeutic agent, a comparison was made between the distribution of the percentage of relative tablet potency [(tab¬let assay/tablet weight) x 100] for the two mixing times. The frequency distribu¬tions of the two populations are shown in Figure 3.
The two histograms are visually different, with the 15-min process exhibit¬ing more dispersion. Despite this difference both populations are tightly grouped, which is a reflection of the uniformity of the blend.

The processes may be studied quantitatively by comparing the means and standard deviations of the two populations. The effect of final blend time on lubricant distribution was examined by comparing disintegration time statistics for the grouped data. None was noted.
The moisture content of the 15 tray-dried batches following final mix remained essentially unchanged from the drying step. The batches from the fluid bed process gained moisture. This is probably attributable to handling very dry material in a relatively humid environment. Both groups are still below the target for this step of 1.5 %, however.
Table 2 gives a comparison of the moisture contents following the drying and tumbling steps. The sizable increase in mean moisture content of the fluid bed-dried batches deserves further study. To determine whether or not all batches were uniformly affected, the mean moisture content was plotted in the order in which the batches were produced. Whereas the plot for the tray-dried batches is unremarkable, the fluid bed process chart (Fig. 4) depicts an unnatural pattern. Further investigation discloses that heating, ventilation, and air condi¬tion (HVAC) problems were experienced by the area in which a number of these batches were blended.
- During compression, 1000 tablets were randomly selected for use by qual¬ity control. Inspection of the batch records revealed that all 30 batches were compressed on the same model press operating at approximately the same speed. All presses were fed by overhead delivery systems of the same design, thus tableting equipment will not be a source of variability from batch to batch.
The test for disintegration is performed as described in the USP, and the results are rounded to the nearest half-min. Disintegration time varied over a narrow range for all batches studied. The 15-batch average for the tray dryer process (2.7 min) is well below the specification (10 min) for this test. Hardness of tablets from the tray dryer process averaged 15 Strong-Cobb units (SCU).
On average, tablets from the fluid bed process were slightly harder. Also, the individual batches had a greater range of hardness than batches from the alternative drying process. Disintegration time for the fluid bed process averaged 3.0 min. Individual batches ranged from 2.0 to 4.5 min. As with the tray process, no correlation was found between hardness and disintegration time. In summary, tablets from the fluid bed dryer process were somewhat harder and took slightly longer to disintegrate. (See Table 2.) These differences are considered insignifi¬cant, however. If any recommendations were made, it would be to lower the disintegration time specification or establish an internal action limit closer to the historical upper range of the process.
Control charts were plotted for hardness and average tablet weight (ATW) to evaluate process performance over time. Separate charts were prepared for the tray dryer and fluid bed processes. Hardness values are an average of 10 individual measurements. The ATW subgroups are the result of weighing 20 tablets individually. The control charts were inspected for trends and evidence of instability using well-established methods [9]. Only the control chart for hard¬ness of tablets from the fluid bed process responded to one of the tests for pattern instability (Fig. 5); that is, two of three consecutive points exceeded the 2-sigma limit. From the chart it is obvious the general trend toward greater tablet hardness (from 11 to 25 SCU) is the underlying cause of the instability. The trend to greater hardness was subsequently arrested and may have to do with attempts to regulate another tablet variable—thickness, for example— although the records are vague in this regard.
Water content of the bulk tablets irrespective of the drying process was higher than at the final mix stage (Table 2). This is probably due to the compres¬sion room environment and the low initial moisture of the powder. Still, the specification limit of 2% is easily met.
The FDA has recently issued draft guidelines that recommend blend uni¬formity analysis for all products for which USP requires content uniformity analysis [10]. The USP requires this test when the product contains less than 50 mg of the active ingredient per dosage form or when the active ingredient is less than 50% of the dosage form by weight. The concern FDA has is that if blend uniformity is not achieved with mixing of the final granulation, then some dosage units are likely not to be uniform [11]. Blend uniformity is not routinely determined for drug A, nor is there a requirement because the dosage form is over 50% active ingredient. In the absence of historical information about uni¬formity of the blend, the relationship between tablet weight and potency should be carefully examined.
Tablet weight should bear a direct relationship to milligrams of active ingredient available where the final blend is homogeneous. This conclusion as¬sumes that demixing does not occur as the compound is transferred to intermedi¬ate storage containers or to a tablet press hopper [12]. To measure the likelihood that controlling tablet weight assures dosage uniformity, 50 tablet assays se¬lected at random (from 300 tablet assays) were compared to tablet weight using regression analysis. Because the same model tablet press and blender were em¬ployed for every batch, assay results from all 30 batches were pooled. The mean purity of the 25 receipts of active ingredients used to manufacture the 30 batches in the validation study was 99.7%, or 0.3% below target.
In conclusion, drug A production was shown to be within established specifications, and there is no reason to believe this will not be the case for future production as long as all practices are continued in their present form. Furthermore, there is no significant difference between batches produced by the tray dryer process and the fluid bed process. A validation report should memori¬alize these findings. The report should also recommend eliminating the option to use a no. 5 screen for the wet milling step and a no. 12 screen to pulverize the dried granulation. There is no experience or only limited experience with this equipment that supports its continued availability. In the same vein, the final blend time should be standardized at 10 min and automatically controlled by means of a timer.
B. Coated Tablet (Drug B)
Let's now turn our attention to a different dosage form, applying some of the strategies developed during the examination of drug A. Again we want to iden¬tify the process steps that are responsible for distributing the active ingredient as well as the tests that measure the effectiveness of those actions. Drug В is a sugar-coated tablet prepared in the traditional manner; that is, layers are slowly built up around a core by applying a coat of shellac and then subcoating, grossing, and smoothing coats until specifications are met at each stage. In the case of drug B, the core contains two active ingredients. The coating, on the other hand, has no medicinal value and is intended solely to enhance the aesthetic appearance of the product. The manufacturing process is shown in Figure 6.
Table 3 summarizes the selected critical steps for the manufacture of the core tablet of drug B. The core is prepared by dry-blending the first active ingredient (i.e., Bl) with several excipients. Blend time is of interest for its impact on the distribution of the therapeutic agent. The premix just prepared is granulated using an alcohol-binder solution. The process directions allow the operator some latitude in using additional alcohol to ensure that the batch is uniformly wet. It will be necessary to know whether or not additional alcohol is routinely required, and if so, how much is used. Besides measuring operator technique, the wetting step affects particle size distribution. The oven tray dryer is identified for drying the wet mix. Granulation drying time is of interest, be¬cause loss on drying is not measured. Once dry, the granulation is milled using a specified screen size and machine setting. Alternate equipment is not provided for in the aforementioned steps.
The powder produced in the prior operation is combined with the second active ingredient (B2), as well as several other excipients in a twin-shell blender and mixed for several min. For reasons previously discussed, mix time is of interest, and thus it is listed as a critical process step.
The blend of the two active ingredients (Bl and B2) is slugged and then the slugs are oscillated. Slugger model and tooling are listed in the batch instruc¬tions. The thickness of the slug is specified, but no information is recorded on the slugging operation, as control of this procedure is left to the experience of the press operator. The batch record permits the use of only one screen size. Since all of the batches have been made in the same manner, this important process step will not be included as one to be studied.
Next, lubricant and oscillated granulation are blended for several min. The elapsed mixing time is of interest because of its impact on drug distribution and the effect of the lubricant on dissolution. During compression, 1000 randomly selected cores are accumulated for use by quality control.
The ATW, hardness, and disintegration time are determined by the press operator during compression. As in the case of drug A, we will not rely on these results for our study, but rather on the test data from quality control.
Following approval of the bulk cores by quality control, they are shellac-coated. According to the manufacturing directions, one or two coats may be applied based on the process operator's judgment. A third coat is permissible but only in response to directions from the supervisor. In any event, the actual number of coats applied is recorded in the batch record.
1. Evaluation of Historical Data
Only 19 batches of drug В are available for examination, one shy of the mini¬mum number previously suggested. The obvious course of action is to delay the study until additional batches are produced. For reasons that will become appar¬ent later, the data analysis will be started with the batches immediately available.
Inspection of assembled data for the 19 batches of drug В confirmed that premix blending was consistently performed for 15 min as specified in the man¬ufacturing directions.
On average, 11.5 kg of additional alcohol was needed to wet the premix adequately. The actual quantity used ranged from 6 to 16 kg, and in no instance was a batch produced without the use of extra alcohol. These data support an increase in the minimum quantity of alcohol that is specified in the manufactur¬ing directions.

Granulation drying time was unremarkable. All 19 batches were dried within the specified time of 12 to 16 hr; the mean time was 13.4 hr, and no trends, seasonal or otherwise, were detected.
The operator is instructed to combine the premix containing active ingre¬dient Bl with active ingredient B2 and blend for 30 min. All 19 batches were handled as directed in the batch record. Oscillation of the slugs back to powder was accomplished in every case using the screen listed in the batch record. For final granulation, we found that each batch was blended for 30 min, as directed. There is no blend uniformity testing.
Once the cores are compressed, one to three sealing coats may be applied by the process operator. The third coat was never required, however. All 19 batches were completed with two coats of shellac. The volume of shellac applied was always 350 mL for both steps, as required by the batch record, and the record further indicates that the temperature of the air directed into the coating pan was always set at 40QC. There is no record of the temperature being moni¬tored, however. The shellacked cores were dried overnight at 35°C. The dryer temperature was tracked and automatically recorded; no variablity was encoun¬tered when the temperature chart was reviewed.
The marketable dosage unit is arrived at by the slow buildup of layers on the shellacked core through the hand application of coating solution. This finish¬ing step is intended solely to enhance appearance by concealing surface irregu¬larities and should have no effect on drug delivery. The three coating solutions are compounded as part of the batch process and immediately prior to being needed. The directions call for the subcoating solution to be held at 65°C ± 2° following compounding and applied at this temperature, Up to five applications are permissible to achieve the tablet target weight of 380 mg; however, for the 19 batches in this study either three or four coats were applied. The impact of varying the number of solution applications was studied by forming the batches into two populations. Mean tablet weight, total volume of solution applied, and mean disintegration time were compared. Unfortunately, the only available dis¬integration measurement was from a test run on the fully built-up tablet (Table 4). The tablets from batches with three applications of subcoat solution had slightly lower weights on average (6 mg), relative to the other group. The vol¬ume of coating solution varied considerably by application (475 to 700 mL), and the total volume was slightly lower when there were only three applications. Mean disintegration time of the groups differed by less than 30 sec, which is insignificant, given the test methodology.
Additional layers are added to the tablet using a grossing solution that is similar to the subcoating formula and contains a colorant. As many as 15 appli¬cations may be needed to achieve the target weight of 450 to 490 mg. Warm air (32-38°C) is applied between coats to achieve drying. A dial thermometer is visible to the operator, but there is no requirement to log the actual tempera-

Drug B: Comparison of Mean Hardness and Distintegration Times

Disintegration time (min)
Batch Hardness Core Shellacked Coated
number (SC units) tablets cores tablets
01 11 8 15 22
02 10 9 20 25
03 10 8 19 21
04 11 9 16 22
05 8 8 13 17
06 8 8 14 18
07 8 7 14 21
08 9 8 14 20
09 8 8 15 20
10 10 8 17 19
11 12 9 13 20
12 12 8 13 20
13 8 7 14 17
14 8 7 13 18
15 12 8 13 18
16 10 11 17 23
17 11 9 20 26
18 10 8 18 20
19 9 7 14 19
% 9.74 8.16 15.37 20.32
RSD 15.30 11.76 15.81 12.15
ture. In the manner previously discussed, the total number of applications, vol¬ume of solution consumed, and tablet weight achieved were analyzed. Variabil¬ity was present between batches, but populations that received different treatment were quite similar with respect to tablet weight and disintegration time (as mea¬sured at the finished tablet stage).
A finishing solution is used to bring the tablet to its final weight. The operation is very much as previously described except that fewer coats are ap¬plied and therefore less weight is added. An analysis of the data would follow the strategy just discussed.
Let's next direct our attention to the testing done by quality control. The ATW at the core stage is based on the results from weighing 20 randomly selected tablets. The control chart in Figure 7 depicts a process with no single Granulation drying time was unremarkable. All 19 batches were dried within the specified time of 12 to 16 hr; the mean time was 13.4 hr, and no trends, seasonal or otherwise, were detected.
The operator is instructed to combine the premix containing active ingre¬dient Bl with active ingredient B2 and blend for 30 min. All 19 batches were handled as directed in the batch record. Oscillation of the slugs back to powder was accomplished in every case using the screen listed in the batch record. For final granulation, we found that each batch was blended for 30 min, as directed. There is no blend uniformity testing.
Once the cores are compressed, one to three sealing coats may be applied by the process operator. The third coat was never required, however. All 19 batches were completed with two coats of shellac. The volume of shellac applied was always 350 mL for both steps, as required by the batch record, and the record further indicates that the temperature of the air directed into the coating pan was always set at 40°C. There is no record of the temperature being moni¬tored, however. The shellacked cores were dried overnight at 35°C. The dryer temperature was tracked and automatically recorded; no variablity was encoun¬tered when the temperature chart was reviewed.
The marketable dosage unit is arrived at by the slow buildup of layers on the shellacked core through the hand application of coating solution. This finish¬ing step is intended solely to enhance appearance by concealing surface irregu¬larities and should have no effect on drug delivery. The three coating solutions are compounded as part of the batch process and immediately prior to being needed. The directions call for the subcoating solution to be held at 65°C ± 2° following compounding and applied at this temperature. Up to five applications are permissible to achieve the tablet target weight of 380 mg; however, for the 19 batches in this study either three or four coats were applied. The impact of varying the number of solution applications was studied by forming the batches into two populations. Mean tablet weight, total volume of solution applied, and mean disintegration time were compared. Unfortunately, the only available dis¬integration measurement was from a test run on the fully built-up tablet (Table 4). The tablets from batches with three applications of subcoat solution had slightly lower weights on average (6 mg), relative to the other group. The vol¬ume of coating solution varied considerably by application (475 to 700 mL), and the total volume was slightly lower when there were only three applications. Mean disintegration time of the groups differed by less than 30 sec, which is insignificant, given the test methodology.
Additional layers are added to the tablet using a grossing solution that is similar to the subcoating formula and contains a colorant. As many as 15 appli¬cations may be needed to achieve the target weight of 450 to 490 mg. Warm air (32-38°C) is applied between coats to achieve drying. A dial thermometer is visible to the operator, but there is no requirement to log the actual tempera-

Table 4 Drug В: Comparison of Mean Hardness and Distintegration Times

Disintegration time (min)
Batch Hardness Core Shellacked Coated
number (SC units) tablets cores tablets
01 11 8 15 22
02 10 9 20 25
03 10 8 19 21
04 11 9 16 22
05 8 8 13 17
06 8 8 14 18
07 8 7 14 21
08 9 8 14 20
09 8 8 15 20
10 10 8 17 19
11 12 9 13 20
12 12 8 13 20
13 8 7 14 17
14 8 7 13 18
15 12 8 13 18
16 10 11 17 23
17 11 9 20 26
18 10 8 18 20
19 9 "7 14 19
X 9.74 8.16 15.37 20.32
RSD 15.30 11.76 15.81 12.15
ture. In the manner previously discussed, the total number of applications, vol¬ume of solution consumed, and tablet weight achieved were analyzed. Variabil¬ity was present between batches, but populations that received different treatment were quite similar with respect to tablet weight and disintegration time (as mea¬sured at the finished tablet stage).
A finishing solution is used to bring the tablet to its final weight. The operation is very much as previously described except that fewer coats are ap¬plied and therefore less weight is added. An analysis of the data would follow the strategy just discussed.
Let's next direct our attention to the testing done by quality control. The ATW at the core stage is based on the results from weighing 20 randomly selected tablets. The control chart in Figure 7 depicts a process with no single relationship was found between core hardness and uncoated core disintegration time. The 5-min increase in mean disintegration time from shellac coated core to finished tablet is a measure of the contribution made by the finishing steps.
Receipts of active ingredient raw materials Bl and B2 are accepted by quality control based on standard tests for potency, chemical attributes, and particle size. Particle size is determined by sieve analysis. Unfortunately, this is a limit test in which 99% of the sample must pass through a certain mesh screen, therefore any influence particle size distribution might have on dosage form potency cannot be examined.
Figure 8 is a plot of mean assay results for active ingredient Bl. Drug potency (200 mg per tablet) is measured in duplicate from samples obtained by grinding a composite of 20 randomly selected tablets. Figure 8 is also influenced by the variability of the purity of the raw material, which ranged from 97.6-99.5%.
Nevertheless, the pattern was unresponsive to our standard tests for pro¬cess instability, and individual batch results were well within established control limits for this product (180 to 220 mg). The grand mean of 99.0% is 2.0 mg below the theoretical tablet potency, probably because of below-target purity of the active ingredient raw material.
Content uniformity testing is not a requirement for drug substance Bl, hence no information is available about the weight of the active ingredient in individual dosage units. With so much emphasis today on demonstrating ade¬quate control over this variable, a one-time study run concurrently with the next production should be considered. Kieffer and Torbeck suggest two statistical value outside the upper control limit (UCL) or the lower control limit (LCL). Other tests for instability show the process to be operating normally. All 19 batches were compressed on the same model press, according to the batch re¬cord. The ATW for the coated tablet is shown for comparison. Correlation be¬tween core weight and finished tablet weight is poor. Such fluctuations would be expected of a manual coating operation intended solely to enhance pharma¬ceutical elegance, nevertheless the control chart did not respond to our tests for patterns of instability (Fig. 7).
Disintegration time is measured at three steps in the process: at compres¬sion, after application of the second shellac coat, and at finished product release. Table 4 compares the values of mean hardness obtained for 10 individual cores to the disintegration times for the core, shellacked core, and coated tablets.
techniques—the tolerance interval and capability index (Cpk)—may be used to demonstrate uniformity of the drug substance in the dosage form [14]. The starting point is to assay individually a representative sample of tablets (e.g., 30} from a series of batches. Regression analysis also can be performed to assess the influence of tablet weight and raw material purity on potency with the avail¬ability of data for individual tablets.
Active ingredient B2 (25 mg per tablet) is measured on 10 individual tablets per batch. We randomly selected 50 tablets from the 19 batches for use in regression analysis. Because purity of the raw material varied from 98.4-99.7%, it was included as the second variable. Our predictor equation for tablet potency (y) is
у = -51.10 + 0.5342Xj + 0.0752X2
Xj = raw material X2 = tablet weight
The slope of the regression plane was found to be positive for both tablet weight and raw material purity, as we would expect. The slope for tablet weight was statistically significantly different from 0 at a = 0.01, while the slope for purity was significant at a = 0.05.
Substituting the ideal tablet weight (at the core stage) of 320 mg and mean raw material purity of 99% in the above equation yielded a tablet potency of 25.85 mg, or 0.85 mg greater than theoretical. The predicted tablet potency is close to the ideal and well within specification limits (22.5 to 27.5 mg). It is possible this outcome was influenced by differences arising from the method of determining the purity of the raw material and the potency of the dosage form. The former is a wet chemistry analysis, whereas the potency of the drug in the finished tablet is determined by use of an automated procedure. Unfortunately, we were unable to quantify this difference.
The process for drug В has been shown to operate within narrow limits and yield finished dosage forms that are therapeutically equivalent, as measured by standard product release criteria. There is no reason to believe subsequent batches will perform differently as long as all conditions remain static. Despite this generally favorable prognosis, additional work is necessary to provide the assurance of process reliability expected today.
1. There remains the unanswered requirement to demonstrate blend uni¬formity of active ingredients B2. This issue might be addressed by testing the blends of a series of batches until sufficient data are accu¬mulated to consider the process reliable. Hwang et al. have provided some insight into establishing an in-process blend test [15]. The validation committee might also suggest that an individual tablet assay be performed for active ingredient В1 during this period. The afore¬mentioned statistical treatments would then be employed to demon¬strate that tablet potency is well controlled.
2. Only 19 batches of drug В were considered suitable for the validation study. This number is shy of our stated goal of a minimum of 20 batches. We therefore will want to supplement the data from the original 19 batches. This effort should be coordinated with the blend uniformity testing.
3. Details of the slugging step need to be improved, both to assure consistency and to facilitate third party monitoring. All of these recommendations should be memorialized in the validation report.
С Softgeis (Soft Gelatin Capsules,- Drug C)
This dosage form consists of a solution of active ingredient encased within a spherical, plasticized gelatin shell. Unlike hard gelatin capsules, for which sev¬eral discrete operations are required to produce the final product, the softgel is formed, filled, and hermetically sealed in one continuous operation [16]. Molten gelatin mass is formed into two sheets or ribbons, each of which passes over a die of the desired size and shape. At the point at which the two rotating dies meet, the hemispheres are sealed and simultaneously filled with the solution of active ingredient. Next the capsules are cleaned by immersion in an organic solvent, dried, and inspected. (See Fig. 9.)
According to the process instructions, the active ingredient powder is dis¬solved in vegetable oil with the aid of a solubilizer. Blend time is stated as 25 to 30 min. This is an elapsed time. Because a range of time is permitted, this step is one for which historical data will be sought (Table 5). The bulk solution is assayed to confirm that the prescribed weight of drug С was charged and dissolution is complete before capsule filling may proceed. Concentration of the active ingredient should vary very little from one batch to another with such a straightforward process. We will want to confirm that this is the case. The purity of each active ingredient raw material receipt is also of interest for reasons previously stated.
The instructions for gelatin mass preparation direct that gelatin powder be blended with water, a plasticizer, and colorant until a uniform consistency is achieved, then heated until molten. The recommended blend time is 20 min at a temperature of 60°C + 5°. The temperature of the molten gelatin just prior to formation into a ribbon is critical; too high a temperature causes the gelatin to deteriorate, and a low temperature affects flow rate. Both conditions are to be avoided for their deleterious effect on capsule formation. For these reasons, as follows: die rotation speed controls dwell time. If there is insufficient contact time, the capsule halves will not properly seal. Subpotent softgels may result from loss of liquid fill through a poorly developed seam. Gelatin ribbon thick¬ness determines capsule wall and seam thickness. Insufficient thickness will contribute to poorly formed capsules and leakers. An overly thick ribbon results in shell sealing problems. Ribbon condition is influenced by the temperature of the gelatin mass, as previously noted. Relative humidity in the encapsulation room is important to efficient drying. Minimally, we will want to know the room condition during the time in which the 20 batches in this study were manu¬factured. It would be best to examine environmental conditions over a longer time period, say 1 year, to capture seasonal trends should they exist.
The batch record instructs the encapsulation machine operator to measure and record seam and wall thickness every 45 min. Softgel weight is also checked periodically by this operator. This information could be useful in demonstrating process control but to a large extent seam and wall thickness are controlled by manufacturing conditions for which historical data are already being sought. For this reason, the results of these in-process monitors need not be pursued initially. Consistent with the approach taken for other dosage forms previously discussed, finished softgel weight data can be obtained from quality control reports when dissolution and assay results are collected.

1. Evaluation of Historical Data
The first step in the production sequence is solubilizing the active ingredient in an appropriate volume of vehicle. For drug C, this blend is a solution, and the activity was routinely accomplished in the prescribed time (25 to 30 min). The analytical test results of each bulk batch confirmed that small differences in mix time had no impact. The nine receipts of active ingredient raw material used to prepare the 20 batches under review had a mean potency of 99.5%. Individual receipts ranged from 98.7-102%. No trends were noted when these receipts were examined graphically.
Gelatin mass preparation time was recorded as being between 17 and 23 min. Such small differences were not thought to be worthy of further consider¬ation. Gelatin mass temperature is critical for reasons previously noted. The temperature range achieved during compounding was examined by means of the recorder charts for evidence of equipment problems and lack of operator atten¬tion. The degree of variability within a batch and from batch to batch was considered reasonable for an operator-controlled process of this type. Mass tem¬perature at the end of compounding, just before the start of encapsulation, aver¬aged 60.5°C. Individually, all batches met the specifications of 60°C ± 5°. Con¬trol over gelatin mass temperature for the duration of the filling operation was generally unremarkable, although larger fluctuations were present for four of A bloom strength determination is part of the acceptance criteria for each receipt of gelatin raw material. The bloom gelometer numbers range from 125 to 195 for the 12 lots, with a mean of 147. This number was compared, to gelatin ribbon thickness and die rotation speed during encapsulation to ascertain whether lot-to-lot differences had to be compensated for. No relationship was
Encapsulation machine setup specifications were considered for their im¬pact on softgel seam and wall formation. Die speed is given as 4.0 rpm + 0.2. Gelatin ribbon thickness is to be controlled at 0.032 in. ± 0.003. More than one machine was used to produce the 20 batches; however, they were all the same make and model. Machine settings during encapsulation are summarized in Ta¬ble 6. Slight machine-to-machine differences are present, but all three operations are easily within suggested settings for this product. On average, gelatin mass temperature was the same for each encapsulation machine.
The influence of gelatin mass temperature, gelatin ribbon thickness, and die speed on softgel formation and the interactions of these variables were ex¬plored by regression analysis as follows:
Finished softgel weight = gelatin mass temperature + die speed + gelatin ribbon thickness
The outcome was inconclusive, probably due in part to use of data that did not take into consideration the variability in fill volume.
Quality control release testing was performed on a sample taken from 1000 softgels randomly selected at the conclusion of processing. The outcome of dissolution, assay, and average fill weight tests is reported in Table 7, along with the corresponding specification. These data were analyzed using methods previously illustrated. In addition, all batches passed the microbial limits test.

Dissolution and average fill weight results are not remarkable. Active in¬gredient assays averaged 16 mg above midpoint of the specification, which is not assignable to raw material purity which averaged 99.5%. Examination of in-process checks of wall thickness showed this parameter to be under control at all times, effectively ruling out fill volume as a factor. One explanation could be the manner in which the active ingredient solution is prepared. It is notewor¬thy that all 20 batches exceed the midpoint of the bulk solution specification. Individual batches range from 509 to 523 mg when expressed in terms of target fill weight (900 mg). This distribution suggests that a condition common to all the batches is part of the explanation. The analytical methodology used to re¬lease the bulk and finished dosage form would be a good place to start such an investigation.
Available information reveals a process that is consistently reproducible and can be considered validated on that basis. Before doing so, however, the assay results should be justified and the outcome of this investigation included in the validation report.
The solution dosage form to be discussed is an elixir. A review of the batch record shows that it contains two active ingredients (Dl and D2). The different steps in preparing the dosage form are outlined in Figure 10.
Drug D may be produced in both 1000- and 2000-gal batches to meet inventory requirements. Major equipment and operator instructions are the same regardless of batch size. The only difference is the amount of each ingredient charged to the make tank. With a formulation such as this, there is little likeli¬hood that batch size is an important process variable. Nevertheless, we will be conservative and treat each size batch as a unique process. An alternative strat¬egy would be to validate the 2000-gal process and demonstrate for the 1000-gal batch the adequacy of mixing, using, for instance, assay data.
The batch is prepared using a single tank. Large-volume liquid excipients and deionized water are metered into the main tank. The other materials are


Вернуться в «Валидация очистки»