B. Asbestos Filters

We revised Sec. Sec. 210.3(b)(6) and 211.72 to eliminate provisions permitting limited use of asbestoscontaining filters used in processing injectable drug products. We had proposed to simply delete references to asbestos filters in these provisions. However, in response to comments, we also added to Sec. 211.72 the statement ``The use of an asbestoscontaining filter is prohibited.'' Also in response to comments, we revised Sec. 211.72 to reflect appropriate technical standards for nonfiberreleasing filters.

C. Verification by a Second Individual

The final rule makes several changes to the regulations to acknowledge, consistent with our longstanding interpretation, that certain operations may be performed by automated equipment and verified by a person, rather than one person performing an operation and another person verifying that the operation was correctly performed. In particular, we added new paragraph (c) to Sec. 211.68 stating that automated equipment used to perform operations addressed in Sec. Sec. 211.101(c) or (d), 211.103, 211.182, or 211.188(b)(11) can satisfy the requirements in those sections for the performance of an operation by one person and checking by another person if the equipment is used in conformity with Sec. 211.68 and one person checks that the operations are properly performed. In response to comments, we revised the paragraph to minimize the possibility that the provision might be misinterpreted as requiring a person to repeat by hand all calculations performed by automated equipment.

In accordance with the addition of Sec. 211.68(c), we are adopting corresponding changes to the following provisions:

D. Other Minor Changes

In addition to the revisions to the regulations previously noted, we have made minor revisions to the following provisions to provide greater clarity without changing meaning or intent:

III. Comments on the Proposed Rule and FDA's Response

We received comments on the proposed rule from drug and biologic manufacturers, industry associations, consultants, and other interested persons. A summary of the comments received and our responses follow. We first respond to comments of a general nature and then to comments on the five topics set forth in the preamble of the direct final rule.

To make it easier to identify comments and our responses, the word ``Comment,'' in parentheses, appears before the comment's description, and the word ``Response,'' in parentheses, appears before our response. We have numbered each comment to help distinguish between different comments. Similar comments are grouped together under the same number if the same response would be given for each. The number assigned to each comment is purely for organizational purposes and does not signify the comment's value or importance or the order in which it was received.
A. General Comments
(Comment 1) One comment stated that it will be very important for FDA to ensure clarity and consistency in the understanding of the final rule among agency staff, including both product reviewers and CGMP inspectors, to minimize different interpretations and applications of these regulations.
(Response) We agree that it is important that FDA employees who perform application reviews, as well as conduct CGMP inspections and other compliance activities, understand these regulations and apply them in a consistent manner in the performance of their duties. Therefore, we will take appropriate steps to ensure that agency staff receive adequate training regarding the new regulations.
(Comment 2) One comment stated that we should not withdraw the 1996 proposed rule because it contained many good features with respect to test method validation and the outofspecification test result problem. The comment maintained that the guidance for industry entitled ``Investigating OutofSpecification (OOS) Test Results for
Pharmaceutical Production'' (71 FR 60158, October 12, 2006) is not helpful to people working with biological drugs and other products. Another comment stated that the December 2007 proposed rule should have incorporated many of the changes in the 1996 proposed rule regarding such matters as validation, quality control unit responsibilities, batch failure investigations, and stability samples because they involve some of the most common CGMP deficiencies.
(Response) As we stated in the December 4, 2007, document, we withdrew the 1996 proposed rule because we concluded that, given our new approach to CGMP under the 21st century initiative, it would be preferable to revise the CGMP regulations incrementally rather than in a onetime, comprehensive fashion. Furthermore, we believe that it is appropriate to reevaluate some of the matters considered in the 1996 proposed rule in light of recent scientific and technological advances. We appreciate the comments' interest in the specified CGMP issues, and we will consider these issues in future phases of our CGMP modernization efforts.
(Comment 3) One comment encouraged FDA to consider other CGMP regulations that need modernization or clarification, or are no longer necessary due to technological advances, such as aspects of 21 CFR 610.12 concerning the requirements for bulk sterility testing and allowance for sterility retesting for biological products.
(Response) We appreciate the comment's interest in modernizing CGMP regulations. As previously stated, this final rule represents only our first step in updating the drug CGMP regulations to reflect current industry practice and harmonize the regulations with international CGMP requirements. We will consider other aspects of CGMP in future rulemaking proceedings.

B. Plumbing

Section 211.48(a) requires that potable water be supplied under continuous positive pressure in a plumbing system free of defects that could contribute contamination to any drug product. It further requires that potable water meet the standards established by the U.S. Environmental Protection Agency (EPA) for primary drinking water in 40 CFR part 141. Proposed Sec. 211.48(a) would have deleted the requirement that the potable water used in a plumbing system meet EPA's standards for primary drinking water, and instead required that the water be ``safe for human consumption.'' This proposed revision was intended to improve harmonization with foreign regulations (particularly those of the EU and Japan) and to make the U.S. regulation more consistent with the United States Pharmacopeia standard. In the preamble of the direct final rule, we stated that the revised requirement could be met by compliance with the standards in the EPA regulations or in the current regulations of the EU or Japan for potable water used to prepare water for pharmaceutical purposes. (Comment 4) Four comments objected to the proposed change. Among other things, the comments stated that the standard of ``safe for human consumption'' is not sufficiently prescriptive.
(Response) Because of the comments received and other
considerations, we have decided not to revise Sec. 211.48(a) at this time. We will address the issue of standards for water used in a facility's plumbing system when we consider proposing regulations for water used as a drug product component in the next phase of our CGMP initiative.

C. Aseptic Processing

In the proposed rule, we sought to amend several regulations on aseptic processing to reflect current industry standards and practices. Some of the proposed revisions would also affect other types of processes and operations. We noted that the proposed changes would not affect the applicability of the guidance for industry entitled ``Sterile Drug Products Produced by Aseptic ProcessingCurrent Good Manufacturing Practice'' (Aseptic Processing Guidance), issued on October 4, 2004 (69 FR 59258).

1. Equipment Cleaning and Maintenance (Sec. 211.67(a))

The version of Sec. 211.67(a) amended by this final rule stated: ``Equipment and utensils shall be cleaned, maintained, and sanitized at appropriate intervals to prevent malfunctions or contamination that would alter the safety, identity, strength, quality, or purity of the drug product beyond the official or other established requirements.'' We proposed to add the phrase ``and/or sterilized'' after the word ``sanitized'' in Sec. 211.67(a) to reflect the fact that sterilization is appropriate for sterile drug products.

On our own initiative, we have revised Sec. 211.67(a) to state that equipment and utensils shall be cleaned, maintained, ``and, as appropriate for the nature of the drug, sanitized and/or sterilized at appropriate intervals * * *.'' This revision does not alter the meaning of the proposed rule change, but clarifies that for some equipment and utensils
[[Page 51922]]
used in the production of certain drug products, sanitization is appropriate; for other equipment and utensils, sterilization is appropriate; and for still others, both sanitization and sterilization are appropriate.
(Comment 5) One comment stated that it is not appropriate to address sterilization in Sec. 211.67(a). Instead, the comment recommended that a reference to sterilization of equipment and utensils be added to Sec. 211.113(b), which requires the adoption of written procedures designed to prevent microbiological contamination of drug products purporting to be sterile.
(Response) We do not agree with the comment because, as previously noted, equipment and utensils used in the production of sterile drug products must be sterilized, not merely sanitized. In addition, we have revised Sec. 211.113(b) as discussed in section III.C.5 of this final rule.
(Comment 6) One comment suggested that we could simplify the language in this regulation by changing the phrase ``beyond the official or other established requirements'' to ``beyond the established (or other official) requirements.''
(Response) We do not believe that the suggested change simplifies the current phrase, which we believe is clear. Therefore, we do not believe that the suggested change is necessary.
(Comment 7) One comment stated that Sec. 211.67(a) should not apply to the production of medical gases because most medical gas manufacturing lines are productspecific, closed systems that are not subject to cleaning or sanitation as part of an established periodic cycle, but instead are specially cleaned to be ``oxygen ready'' and carefully handled in accordance with established procedures. The comment maintained that additional cleaning efforts beyond the initial cleaning regimen substantially increase the risk of introducing contaminants into the system. Therefore, the comment stated, it is not necessary to require cleaning of equipment at ``appropriate intervals'' for medical gas manufacturing. The comment suggested that, alternatively, it might be appropriate for the agency to state that medical gases may represent unique circumstances that will be reflected in a separate guidance.
(Response) We decline to exempt medical gases from the requirements of Sec. 211.67(a) as recommended because this would exceed the scope of our proposed change to clarify that sterilization is appropriate for sterile drug products and would instead focus on whether there is any need for periodic cleaning of medical gas systems. We might consider in a future CGMP rulemaking whether it is appropriate to revise Sec. 211.67(a) to address its application to medical gases.
2. Microbiological Testing of Objectionable Lots of Components, Drug Product Containers, and Closures (Sec. 211.84(d)(6))

The version of Sec. 211.84(d)(6) amended by this final rule stated: ``Each lot of a component, drug product container, or closure that is liable to microbiological contamination that is objectionable in view of its intended use shall be subjected to microbiological tests before use.'' We proposed to change the phrase ``that is liable to microbiological contamination'' to ``with potential for microbiological contamination.''
(Comment 8) One comment stated that the proposed change was unnecessarily restrictive and might lead to testing every lot when the risk of microbial contamination is low and the impact on the intended use is insignificant. This comment suggested replacing ``that is liable to microbial contamination'' with ``prone to microbial contamination.'' One comment stated that the proposed change could make it more difficult for drug manufacturers to replace a less effective, quality controlbased inspection and test method with a more modern and effective quality audit method. The comment stated that because the bioburden of dry items such as vials and stoppers is often heterogeneous, improved assurance of this quality attribute is better achieved through the audit, selection, and control by the manufacturers of these items. This comment maintained that knowledge of and control over the manufacturing processes for containers and closures might fall short of justifying that those products do not have a ``potential for contamination.''
(Response) We decline to adopt the recommended change to Sec. 211.84(d)(6) from ``that is liable to microbial contamination'' to ``prone to microbiological contamination.'' We believe that our proposed change to ``with potential for microbiological contamination'' clarifies our longstanding interpretation of the regulation that each lot of component, drug product container, or closure that is susceptible to contamination must undergo microbiological testing before use. Therefore, we have revised Sec. 211.84(d)(6) to refer to components, containers, or closures ``with potential for
microbiological contamination'' as proposed.
3. Validation of Depyrogenation of Drug Product Containers and Closures (Sec. 211.94(c))

The version of Sec. 211.94(c) amended by this final rule stated: ``Drug product containers and closures shall be clean and, where indicated by the nature of the drug, sterilized and processed to remove pyrogenic properties to assure that they are suitable for their intended use.'' In the preamble to the direct final rule, we stated that it has been longstanding industry practice to validate the sterilization and depyrogenation processes used for drug product containers and closures to ensure consistent removal of microbial contamination and pyrogens or endotoxins. Therefore, we proposed to add a provision to Sec. 211.94(c) requiring the validation of these depyrogenation processes.
(Comment 9) One comment suggested that we require validation of ``sterilization'' as well as depyrogenation processes.
(Response) We do not believe that the suggested change is needed because Sec. 211.113(b) already requires validation of sterilization processes for the prevention of microbiological contamination of drug products purporting to be sterile.
(Comment 10) Four comments objected to the requirement in existing Sec. 211.94(c) because it requires depyrogenation of components based on the nature of the drug and does not take into account the fact that some containers and closures are inherently nonpyrogenic, have been qualified not to require active depyrogenation, or do not require depyrogenation because of handling procedures. Three of the comments proposed that in addition to the nature of the drug, the drug's manufacturing process be included as a factor in determining when containers and closures must be sterilized and processed to remove pyrogenic properties. Two of the comments recommended that the requirement to validate depyrogenation processes be limited to containers and closures that are made nonpyrogenic by a designated depyrogenation process (thus excluding inherently nonpyrogenic containers and closures from the regulation).
(Response) We decline to adopt the suggested revisions because they go beyond the scope of our proposed change to require validation of depyrogenation processes and instead focus on the need for depyrogenation itself.
[[Page 51923]]
4. Inclusion of Bioburden Testing in InProcess Testing (Sec. 211.110(a))

Section 211.110(a) requires that written procedures be established and followed that describe inprocess controls and tests or examinations to be conducted on samples of inprocess materials of each batch of a drug product. The regulation specifies five control procedures that must be established, where appropriate, to monitor the output and to validate the performance of manufacturing processes that may be responsible for causing variation in the characteristics of in process material and the drug product. We proposed to add bioburden testing to this list (which is not allinclusive) because testing for bioburden is standard industry practice for inprocess materials and drug products that are produced by aseptic processing.
(Comment 11) Three comments objected to the addition of bioburden testing to Sec. 211.110(a). One comment objected to the inclusion of any specific test and suggested that specific tests be addressed in agency guidance. One comment stated that bioburden testing is not conducted at the same time as other tests specified in Sec. 211.110(a) and is not an inprocess test or control because it does not yield immediate results that allow for process adjustment. The comment stated that it would be more appropriate to address bioburden testing in Sec. 211.84. One comment suggested that because Sec. 211.110 covers the sampling and testing of all inprocess materials and drug products, adding bioburden testing as a mandatory control procedure could expand current industry validation procedure and produce diversity among the industry and regulators on the circumstances in which validation of bioburden testing is appropriate.
(Response) We do not agree with the comments. As stated in the direct final rule, testing for bioburden is an important inprocess control, particularly for drug products that are produced through aseptic processing. Section 211.110(a) provides flexibility to manufacturers so that they need only conduct bioburden testing where the testing is appropriate to assure batch uniformity and drug product integrity. We believe that manufacturers understand for which types of drug products, and at what point in the manufacturing process for these drugs, bioburden testing is appropriate. Accordingly, we have added bioburden testing to Sec. 211.110(a).
5. Control of Microbiological Contamination (Sec. 211.113(b))

Section 211.113(b) states that appropriate written procedures, designed to prevent microbiological contamination of drug products purporting to be sterile, must be established and followed. The version of Sec. 211.113(b) amended by this final rule further stated: ``Such procedures shall include validation of any sterilization process.'' We proposed to substitute ``all aseptic and sterilization processes'' for ``any sterilization process.'' As noted in the preamble of the direct final rule, even before we issued the nowreplaced guidance on ``Sterile Drug Products Produced by Aseptic Processing'' in 1987, industry routinely conducted validation studies (often referred to as media fills) that substituted microbiological media for the actual product to demonstrate that its aseptic processes were validated (72 FR 68064 at 68066). The proposed change was intended to clarify existing practice and to harmonize Sec. 211.113 with Annex 1 of the EU CGMPs. (Comment 12) Several comments objected to the proposed change to Sec. 211.113(b) on the basis that aseptic processing cannot be validated. One comment stated that validation of aseptic processing technically cannot be done, although the manufacturer can ensure tight control over the process. One comment stated that aseptic processing simulations demonstrate the capability of a facility, equipment, and operational controls to provide a minimal microbial contamination rate in a single event, but they cannot predict the outcome of a similar process performed at a different time. The comment maintained that to consider aseptic processing to be validated overstates the ability to measure and control the process and could be interpreted as approval to relax the controls necessary for its success. The comment recommended that Sec. 211.113(b) be revised to require validation of ``all sterilization/depyrogenation processes'' and to direct that aseptic processes ``be subjected to periodic assessment to demonstrate the capability of the control strategy to adequately support end product sterility.''

One comment stated that there is currently no means to comply with the proposed requirement to validate aseptic processes. The comment maintained that the microbiological and decontamination methods used in aseptic processing lack the sensitivity, recoverability, and accuracy of the physical and chemical measurement systems normally associated with process validation. The comment further claimed that media fills do not validate aseptic processing because they measure only detectable microorganisms and do not verify that no microorganisms exist. The comment stated that although aseptic processing cannot be validated, a state of control can be established, ensuring that the aseptically produced drug consistently meets its specifications and quality attributes. The comment recommended that rather than validation of aseptic processes, Sec. 211.113(b) require ``a formalized quality risk management and control strategy for aseptic processes to provide assurance of requisite and continued process capability and product quality.''

One comment stated that although media fills can evaluate an aseptic process, they cannot be considered to validate the process. The comment recommended that we either not adopt the proposed requirement to validate aseptic processes or provide more clarity on what is expected for validation of aseptic processes. Similarly, another comment recommended that we not revise Sec. 211.113(b) as proposed unless we clarify that more than media fills are required to validate an aseptic process. The comment stated that a wellcontrolled, robust process is required for aseptic processes and that once a state of control has been established for the process, media fills can be useful in confirming the state of control.
(Response) Although we acknowledge that aseptic process validation does not provide absolute assurance of product sterility, we do not agree that aseptic processes cannot be validated. Validation of aseptic processes, which is a common practice throughout the pharmaceutical industry, means establishing documented evidence that provides a high degree of assurance that a particular process will consistently produce a product meeting its predetermined specifications and quality attributes. Media fills, together with operational controls, environmental controls, and product sterility testing, provide a sufficient level of assurance that drugs purported to be sterile are in fact sterile.
(Comment 13) One comment suggested adding a definition of aseptic processing to part 210.
(Response) We do not believe that it is necessary to define aseptic processing in the regulation. The Aseptic Processing Guidance makes it clear to manufacturers what aseptic processing entails.
[[Page 51924]]
(Comment 14) One comment requested confirmation that it is acceptable to follow the current FDA guidance and use media fills to meet the requirement to validate aseptic processes.
(Response) As stated in the preamble to the direct final rule and reiterated previously in this document, manufacturers can follow the recommendations in the Aseptic Processing Guidance to comply with CGMP requirements for aseptic processing, including validation. However, as with any guidance, the Aseptic Processing Guidance is not binding on industry or the agency, and manufacturers may use an alternative approach to achieve compliance if the approach meets the requirements of the act and FDA regulations.
(Comment 15) One comment sought clarification that the requirement to validate aseptic processing would not inhibit implementation of novel technologies recommended by the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) in the ICH Q8, Q9, and Q10 guidances, or other innovative approaches in these areas.
(Response) We do not believe that the requirement to validate aseptic processing will interfere with the implementation of new technologies either as part of following ICH recommendations or as part of other efforts to meet CGMP requirements. As stated in section I of this document, we have always attempted to balance the need for easily understood minimum CGMP standards with the desire to encourage innovation and the development of improved manufacturing technologies. We are confident that industry can meet the requirement to validate aseptic processing with no adverse impact on technological innovation in drug product manufacturing.

D. Asbestos Filters

As stated in the preamble to the direct final rule, we need to update our regulations on filters used in processing liquid injectable products. The version of Sec. 211.72 amended by this final rule required manufacturers, before using an asbestoscontaining filter, to submit proof to FDA that an alternative nonfiberreleasing filter will, or is likely to, compromise the safety or effectiveness of the product. However, we are not aware that asbestos filters are currently commercially manufactured for pharmaceutical use or are used in drug production, and their use is not considered a good manufacturing practice. Therefore, we proposed to delete the reference to the use of asbestoscontaining filters from Sec. 211.72 and to delete the reference to asbestos filters from the definition of ``nonfiber releasing filter'' in Sec. 210.3(b)(6).
(Comment 16) Two comments stated that the regulations should state that the use of asbestos filters is prohibited. One comment stated that if asbestoscontaining filters are in fact available and the proposed changes were interpreted as permitting their use, this might pose a risk to patients.
(Response) We agree with the comments. Therefore, in addition to deleting the reference to asbestoscontaining filters in Sec. 210.3(b)(6), we have revised the last sentence of Sec. 211.72 to state that the use of an asbestoscontaining filter is prohibited. (Comment 17) One comment recommended that we clarify the second sentence in proposed Sec. 211.72, which stated: ``Fiberreleasing filters may not be used in the manufacture, processing, or packing of these injectable drug products unless it is not possible to manufacture such drug products without the use of such filters.'' The comment recommended that this sentence be revised to state as follows: ``Fiber releasing filters may be used when/where it is not possible to manufacture such drug products without the use of such filters.'' (Response) We agree with this proposed change and have revised Sec. 211.72 accordingly.
(Comment 18) Four comments recommended revising the following provision in proposed Sec. 211.72: ``If use of a fiberreleasing filter is necessary, an additional nonfiberreleasing filter of 0.22 micron maximum mean porosity (0.45 micron if the manufacturing conditions so dictate) shall subsequently be used to reduce the content of particles in the injectable drug product.'' Each of these comments stated that it is technically more accurate to describe a filter in terms of its nominal pore size rating than its mean porosity. One comment stated that the filter pore size standard of 0.22 micron is outdated and should be changed to 0.2 micron.
(Response) These suggested technical changes are consistent with statements in our guidances for industry (e.g., the Aseptic Processing Guidance) concerning filters. Therefore, we have revised Sec. 211.72 to require that if use of a fiberreleasing filter is necessary, an additional nonfiberreleasing filter having a maximum nominal pore size rating of 0.2 micron be used.

E. Verification by a Second Individual

The current CGMP regulations include several provisions requiring that certain activities be performed by one person and checked as specified by a second person.

When we amended the CGMP regulations in 1978, we established Sec. 211.68, which provides that automatic, mechanical, or electronic equipment or other types of equipment, including computers, or related systems that will perform a function satisfactorily, may be used in the manufacture, processing, packing, and holding of a drug product, subject to the following requirements:

In the preamble to the 1978 final rule, we stated that the verification requirements in Sec. 211.101 for chargein of components when automated systems are used would be met if a person verified that the automated system was working properly (43 FR 45014 at 45051, September 29, 1978). Thus, in this situation, the first individual is replaced by a machine or other automated process, and only one person is necessary to verify that the automated system is functioning as intended.

[[Page 51925]]

Because we have received questions about the performance and checking requirements in Sec. Sec. 211.101(c) or (d), 211.103, 211.182, or 211.188(b)(11) when the operations are performed by automated equipment, such as the widespread and increasing use of computercontrolled operations, we proposed to revise these sections. We proposed to amend these regulations to indicate that when automated equipment is used to perform certain operations, only one person is needed to verify that the automated equipment is functioning adequately. Correspondingly, proposed Sec. 211.68(c) stated that automated equipment used for performance of operations addressed by Sec. Sec. 211.101(c) or (d), 211.103, 211.182, or 211.188(b)(11) can satisfy the requirements included in those sections for the performance of an operation by one person and checking by another person if such equipment is used in conformity with Sec. 211.68 and one person verifies that the operations addressed in those sections are performed accurately by such equipment. We stated in the preamble of the direct final rule that these revisions would clarify our longstanding policy that verification by a second individual may not be necessary when automatic equipment is used under Sec. 211.68.
1. General Comments on Verification
(Comment 19) One comment stated that validated, automated systems equipped with real time alarms that do not require any human intervention should not require human verification. Another comment stated that such systems should not require human verification with each use and, when human verification is needed, the level of verification required should be consistent with the level of automation used. Both of these comments maintained that requiring operator verification of automated, validated equipment under Sec. Sec. 211.68(c), 211.101(c)(3) and (d), 211.103, and 211.188(b)(11) might hinder the implementation of process analytical technology (PAT) in the drug industry.
(Response) In the Federal Register of February 12, 1991 (56 FR 5671) (the 1991 proposal), we issued a proposed rule in part to amend Sec. 211.68 to add what is now the third sentence of Sec. 211.68(b): ``The degree and frequency of input/output verification shall be based on the complexity and reliability of the computer or related system.'' This revision was adopted as part of the final rule issued on January 20, 1995 (60 FR 4087) (the 1995 final rule).

In the 1995 final rule, we responded to several comments on the proposed revision. Two comments suggested that the revised regulation did not accommodate the accepted use of validated computerized drug production and control systems. We declined to change the revision as proposed, stating our belief that the wording in the revised rule adequately encompasses the use of these systems (60 FR 4087 at 4089).

Two comments on the 1991 proposal questioned the need for human verification of operations that are performed by validated computer systems. The comments listed other regulations that were not the subject of the proposed rule that required more than one person to verify certain manufacturing operations, apparently to show that additional personnel would be needed to comply with proposed Sec. 211.68. We noted in the 1995 final rule that the revisions to Sec. 211.68 do not impose any specific personnel requirements. We also noted that the agency is aware that computers are subject to malfunctions, some of which could possibly result in the loss of critical information regarding the manufacturing process or a serious production error and the possible distribution of an adulterated product. Therefore, we stated that while increasingly sophisticated system safeguards and computerized monitoring of essential equipment and programs help protect data, no automated system exists that can completely substitute for human oversight and supervision. We further indicated that while the degree of verification is left to the manufacturer's discretion, the exercise of such discretion under Sec. 211.68 requires the use of routine accuracy checks to provide a high degree of assurance that input to and output from a computer or related system are reliable and accurate. We stated our intent that each manufacturer exercise reasonable judgment based on a variety of factors, including, but not limited to, the complexity of the computer or related system, in developing a method to prevent inaccurate data input and output (60 FR 4087 at 4089).

The December 4, 2007, direct final rule and companion proposed rule were intended to amend the regulations involving secondperson checks only to clarify our longstanding policy that verification by a second individual may not be necessary when automatic equipment is used under Sec. 211.68, and that in such situations only one person is needed to verify that the automated equipment is functioning adequately. The amendments were not intended to either add to or detract from any existing requirements in this regard, but only to clarify our longstanding interpretation and policy for these requirements. We note that the same basic considerations apply in this regard today as we expressed in the 1995 final rule. Although increasingly sophisticated controls and safeguards have been implemented for some automated systems, our policy has been that some degree of human oversight, supervision, verification, monitoring, or checking is still necessary to verify proper performance as part of assuring the identity, strength, quality, and purity of drug products. For suitably validated automated systems, even with real time alarms, it is still necessary for a human to verify that the systems are operating as planned and to monitor for abnormalities. We agree that the level, nature, and frequency of such human verification will vary depending on the level of automation used as well as the nature of the system and controls, and the manufacturer has the flexibility and responsibility to determine what is suitable and necessary. Contrary to the comments, we believe that manufacturers can conduct human verification of automated operations in conjunction with the use of PAT in drug production.

For these reasons, we continue to believe that human verification is necessary to ensure that automated systems are functioning properly. (Comment 20) One comment stated that many current biotech processes include component additions and deletions in a continuous or periodic manner over long periods of time. The comment stated that there would be no added value in requiring a manual verification of this component management scheme in a fully automated scenario.
(Response) For the reasons stated in our response to comment 19, we believe that some degree of human oversight, supervision, verification, monitoring, or checking is a necessary part of CGMP for such processes and that there is added value in having greater assurance that the automated systems are operating properly as intended. We do not expect that each individual component change must be witnessed in person, but rather that a suitable system of human oversight be established and followed to effectively verify that the automated processes are indeed operating correctly in the performance of these operations. (Comment 21) One comment maintained that our statement in the preamble of the direct final rule that the verifying individual may be, but is not required to be, the operator is a
[[Page 51926]]
contradiction of the CGMP regulations, which require (in Sec. 211.25(a)) that all individuals have the education, training, and experience to enable them to perform their assigned functions. The comment asked why the agency would allow an untrained operator to perform a sole verification of a critical step if an automated system is used and recommended that we retract the noted preamble statement. (Response) The comment incorrectly concluded that allowing the verifying individual to be a person other than the operator would thereby allow an untrained individual to perform the function of verifying a critical step. Section 211.25(a) requires each person performing an assigned function to have the education, training, and experience, or any combination thereof, to enable that person to perform the function. Thus, any person, whether the operator or not, who performs such a verification step would necessarily be required to have the knowledge, training, and experience needed to perform that function. Therefore, our preamble statement does not conflict with the regulations.
(Comment 22) One comment stated that the proposed changes regarding second person verification should be extended to include Sec. 211.188(a), which requires the preparation of batch production and control records that include an accurate reproduction of the appropriate master production or control record, checked for accuracy, dated, and signed. The comment stated that when there is only one signature needed, but the system is automated, it would also follow that no human signature or signature equivalent would be necessary, such as in issuance of a batch record under Sec. 211.188(a), when the record is electronic. The comment also stated that in this case, it is impossible to check the pages for a true and accurate copy. The comment recommended revising Sec. 211.68(c) to include Sec. 211.188(a) in the listing of sections affected and to state that there could be single performance verification under Sec. 211.188(a).
(Response) We do not agree with the recommended changes to Sec. 211.188(a), which would eliminate any human verification of the records. As previously stated, we are clarifying in this rule that the checking of automated equipment by one person can satisfy the requirements of those regulations that address the performance of a step by one person and the verification of the step by a second person. Our proposal regarding verification of operations was intended to make clear that only one person is needed to verify that automated equipment for a processing step is functioning properly; we did not propose deleting all human verification of the step. In addition, we disagree with the comment's apparent contention that no human signature would be needed for issuance of electronic batch production and control records. If such records are generated and issued electronically as part of an automated system, a person must verify that the correct records were issued and that they are still accurate and complete. We believe it is clear that Sec. 211.188(a) requires only one check for accuracy, with date and signature (which could be electronic), and that it does not require a separate second check of this step. Therefore, no changes to Sec. 211.188(a) are necessary or appropriate.
(Comment 23) Three comments addressed secondperson verification in Sec. 211.194. Section 211.194(a) requires that laboratory records include complete data derived from all tests necessary to assure compliance with established specifications and standards as specified in that subsection. Section 211.194(a)(7) requires that laboratory records include the initials or signature of the person who performs each test and the date(s) the tests were performed. Section 211.194(a)(8) requires the initials or signature of a second person showing that the original records have been reviewed for accuracy, completeness, and compliance with established standards. Two of the comments stated that the principle behind the proposed secondperson verification revisions should be extended to Sec. 211.194 to include checking laboratory records involving automated laboratory equipment. The first comment recommended revising Sec. 211.194 generally. The second comment specifically recommended that Sec. 211.194(a)(8) be revised to add that if laboratory tests have been performed by automated equipment under Sec. 211.68, the laboratory record need only include the identification of one person conducting the review of the tests performed by the automated system. The comment also asked that Sec. 211.194(a)(8) be added to the list of sections affected in Sec. 211.68(c). The third comment stated that the failure to include Sec. 211.194(a)(7) and (a)(8) in the proposed revisions implies that the use of automated systems to perform or check testing is not allowed. (Response) We decline to include Sec. 211.194 among the sections enumerated in Sec. 211.68(c) concerning secondperson verification of operations performed by automated equipment. We acknowledge that automated equipment may be used to conduct certain laboratory testing operations. However, when automated equipment is used to perform a laboratory test, typically a person initiates the test and ensures that the correct equipment is used and that it operates properly. In this situation, one person assists in or oversees the performance of the laboratory test and a second person reviews the records for accuracy, completeness, and compliance with established standards. Thus, the use of equipment to perform laboratory tests, though permissible, is not a situation in which automated equipment (rather than a person) performs an operation and a person verifies that performance, which is the situation addressed in revised Sec. 211.68(c). Therefore, it would not be appropriate to include a reference to Sec. 211.194 (or to Sec. 211.194(a)(8) specifically) in revised Sec. 211.68(c).
2. Automatic, Mechanical, and Electronic Equipment (Sec. 211.68) (Comment 24) One comment stated that Sec. 211.68 is no longer in line with the technological improvements of the past 30 years and with the increasing knowledge of computer validation by industry and regulators. The comment recommended that Sec. 211.68 be aligned with 21 CFR 820.70(i), section 5.4 of the ICH Q7A guidance entitled ``Good Manufacturing Practice Guidance for Active Pharmaceutical
Ingredients,'' and the Pharmaceutical Inspection Cooperation Scheme's Annex 11 on computerized systems.
(Response) We decline to adopt the suggested revisions because they exceed the scope of our proposed revision of Sec. 211.68, which only addressed secondperson verification of operations performed by automated equipment. We might consider revising other provisions of Sec. 211.68 as part of a future rulemaking to update the CGMP regulations and make them consistent with international CGMP provisions.
(Comment 25) One comment recommended that instead of our proposed changes to Sec. 211.68(c) and other regulations concerning second person verification, we revise Sec. 211.68(a), which permits the use of automatic, mechanical, or electronic equipment in the manufacture, processing, packing, and holding of drug products. The comment stated that the wording of our proposed changes only allows for actions to be performed by automated equipment and checked by a person, which would prevent the introduction of automated systems to check operations performed by a person. The comment also stated that our proposed changes would still require the involvement of at least one person [[Page 51927]]
in each of these circumstances and prevent the use of a controlled system or systems that both perform and independently verify the relevant operations. One comment suggested that rather than our proposed revisions, the desired clarification concerning automated equipment and secondperson checks would be better achieved by adding to Sec. 211.68(a) the following sentence: ``Automated equipment can satisfy the requirements for the performance of an operation by one person and/or checking by another person.''
(Response) We do not agree with the recommended change. The proposed rule simply clarified our longstanding position that only one human check is necessary to verify a processing step performed by automated equipment. The suggested revision of Sec. 211.68(a), however, would allow manufacturers to rely solely on automated equipment to verify the human performance of certain processing steps and allow automated equipment to both perform and check operational steps, which would constitute a significant change from the current regulations. As stated in our response to comment 19, we believe that human verification of certain processing steps, even when those steps are performed by automated equipment, is still necessary.
(Comment 26) One comment stated that although proposed Sec. 211.68(c) implies that the automated equipment is doing the work and a person can verify that the work is done, there are cases in which a person does the work and automated equipment might be able to verify the person's work. The comment cited as an example the case in which an automated system scans the bar codes of ingredients and equipment to ensure that the ingredient is correct for use with the equipment for that step in the process, but the physical addition of the ingredient is by the human operator (followed by the automated system scanning). The comment recommended, therefore, that Sec. 211.68(c) be modified to allow both the automated system and the person to do either the performance or the verification tasks for the operations addressed by Sec. Sec. 211.101(c) or (d), 211.103, 211.182, 211.188(b)(11), or 211.194(a)(8), or a single performance verification in the case of Sec. 211.188(a).
(Response) We acknowledge that it might be possible to design an automated system to verify operations performed by humans, but as stated in our response to comment 19, we continue to believe that some human verification of the processing steps performed by an automated system is necessary.
(Comment 27) One comment suggested revising Sec. 211.68(c) to state that automated equipment can satisfy the requirements for verification of operations addressed by the listed sections as follows: (1) If such unit operation is fully automated, no manual verification is necessary and (2) if there is an operator for the automated equipment, the verifying individual may be, but is not required to be, the operator. The comment gave several reasons for this change:

Section 211.101 concerns chargein of components. Proposed Sec. 211.101(c) stated, in part, that if the weighing, measuring, or subdividing operations for components are performed by automated equipment under Sec. 211.68, only one person is needed to ensure that the requirements in Sec. 211.101(c)(1), (c)(2), and (c)(3) are met. (Comment 30) One comment proposed broadening Sec. 211.101(c) to clarify that the weighing, measuring, and subdividing operations could be either performed by automated equipment or checked by automated equipment after being performed manually.
(Response) We decline to make this suggested change for the reasons provided in response to comments 19 and 25. Revised Sec. 211.101(c) only permits human checking of weighing, measuring, and subdividing operations performed by automated equipment; we did not propose to allow automated checking of these operations. We continue to believe that human verification of these processing steps is necessary. (Comment 31) One comment stated that with respect to medical gases, there is no measurement of components to be dispensed for manufacturing that needs to be doublechecked to ensure that the right quantity of the right component was added, because transfers of pure gases are within productspecific systems. However, the comment stated, with respect to gas mixtures, it is appropriate to have a verification of hookups as different components are added unless there is subsequent purity testing for each component.
(Response) We decline to exempt single gas filling operations from certain requirements of Sec. 211.101(c) as recommended because such a change would exceed the scope of our proposed change to Sec. 211.101(c), which only addressed human checking of weighing, measuring, and subdividing operations performed by automated equipment. We might consider in a future rulemaking whether it is appropriate to exempt medical gases from certain requirements of Sec. 211.101(c). 4. Verification of Components Added to the Batch (Sec. 211.101(d))

Proposed Sec. 211.101(d) would have required that each component be either added to the batch by one person and verified by a second person or, if the components are added by automated equipment under Sec. 211.68, only verified by one person.
(Comment 32) One comment stated that eliminating a double check for adding materials to a batch is problematic because an error in those operations would be difficult to detect and might not be discovered befor

FOR FURTHER INFORMATION CONTACT Mary Malarkey, Center for Biologics Evaluation and Research (HFM600), Food and Drug Administration, 1401 Rockville Pike, Rockville, MD 208521448, 3018276190; or

Dennis Bensley, Center for Veterinary Medicine (HFV140), Food and Drug Administration, 7500 Standish Pl., Rockville, MD 20855, 240276 8268; or

Brian Hasselbalch, Center for Drug Evaluation and Research, Food and Drug Administration, 10903 New Hampshire Ave., rm. 4364, Silver Spring, MD 20993, 3017963279.