GMP Engineering. Is there such a thing? Sure is! The CFR says:
§211.42 Design and construction features.
(a) Any building or buildings used in the manufacture, processing, packing, or holding of a drug product shall be of suitable size, construction and location to facilitate cleaning, maintenance, and proper operations.
It goes on to identify requirements for :
Operations
Lighting
Ventilation
Plumbing
Sewage
Washing and toilet facilities
Sanitation
Maintenance
Equipment design
Equipment construction
Equipment cleaning and maintenance
Automatic, mechanical, and electronic equipment
Filters
Yet in spite of all these requirements, a project engineer in a parenteral plant once challenged a colleague: “You show me where in the regulations Engineering is involved! Engineering is not part of the compliance requirements!”
Wow! Unfortunately this gentleman was too far behind the compliance curve to be redeemed. He had to be “outplaced”.
GMPs start with the engineering function. If Good Manufacturing Practices are not a consideration in the design and building of a facility or installation of equipment, they cannot be added later by procedures or practices. Only additional money spent to remediate the lack of engineering.
The engineering of a new facility or the installation of new equipment should involve End Users, Quality, and Validation folks, yet not to the point where such involvement bogs often fluid deign stages. Nor is the same level of commitment required by the End User, Quality, and Validation for a small project vs. a large new facility.
PTS has developed procedures implementing different levels of review based on project complexity. From dedicated teams to a standing project review team meeting regularly to go over smaller, less complex projects.
In any case, such a review can catch GMP issues while still on papre, rather than on brick and mortar and 316L SS! See www.ptsgmp.com/PTSLifeCycle.pdf for a simple flowchart of the life-cycle of an engineering project.
Proposal for a Maintenance Technician Training/Qualification Program
A. Establish a file for each employee and regular contractor outlining his/her qualifications and training. His file would include a C.V. and any training or certifications received within the company or outside the company. This would include technical training and copies of training records for internal GMP training.
For company employees that do not have a prepared C.V., a form can be developed that would allow the employee to outline his/her training, education, and experience and this can be filed in that person’s file as a starting point.
B. Training
1. Since most PMs or Job Plans are composed of generic maintenance operations, the suggestion is to train technicians on those operations, rather than each PM in depth. For example lubrication, checking belts, checking filters, etc. are operations common to many PMs. Rather than repeat training on lubrication for the PM for AHU-1 and AHU-2 and AHU-3, etc, train once on lubrication and record that training. Thus the maintenance training could be reduced to specific modules that can be modified and/or added as required. Some modules that come to mind would be:
Lubrication
Air Filters - check and/or replace
Checking belts for wear and tension
Fluid handling components
Linkages – check and lubricate
Verifying drive alignment
Electrical maintenance
This list is certainly not comprehensive, but is offered as an indication of possible divisions of maintenance modules. It is anticipated that these module would be very short and could be as short as 30 minutes each, allowing for going through a bunch in a short time.
There may still be the need for some more specialized equipment maintenance training, for example some very specialized equipment maintenance, but the bulk of the training would be simplified and much less redundant.
2. The preceding would be in lieu of training on specific PMs. If training on these was thought to be necessary as an addition it is recommended that the PM training be limited to a Read and Understand basis. The PMs would be compilations of the tasks covered in the Maintenance Module sessions and perhaps specific belt numbers or lubrication specifications required, so a “Reading and Understanding” of the components of the PM as assembled would be all that was required.
This approach would mitigate the training burden and be a more practical approach, training once on the skills required rather than the repetitive training of the same operations for each PM.
A biopharmaceutical company recently discovered a significant deviation in their calibration program and to help others avoid such an issue, the events are recounted here with some analysis of the causes, results and fixes.
A single sentence comment in an assessment of the calibration program of this company initiated the chain of events. The observation noted that the program did not provide accuracy information and requirements for Standards used in the calibration program. This raised the recognition that the Standard Accuracies were not readily available, especially to the techs performing the calibrations.
The administrative SOPs for the program clearly stated that the Standard must be more accurate than the unit under test (UUT). [This statement in itself was a watering down of earlier statements in the Administrative SOPs that the Standard was to be four times as accurate as the UUT, except in approved circumstances. The “four times” statement is a result of a general calibration rule that a valid calibration activity should have a Test Accuracy Ratio of 4:1, but this was reduced to “more than” in the latter SOP.]
In an attempt to provide the accuracies of the Standards to the techs a service company was contracted to survey and provide accuracies for the Standards in use at the site. The initial results of the survey resulted in points to consider and lessons learned:
First, identifying the Standard alone was not adequate. Identification of the Standard and all associated equipment used with the Standard was required to establish the “Standard Train” accuracy. For example, a particular temperature Standard may have an accuracy of ±0.5oC, but that Standard of itself could not measure temperature without a thermocouple, in one case a type J couple with an accuracy statement of ±0.9oC. The accuracy of the Standard Train was therefore ±1.03oC, using the recognized Root of the Sums Squared method.
Lesson 1: The accuracy of all the components of the Standard Train must be known and considered when evaluating the TAR for a calibration.
Second, the list identified circumstances when Standard Trains were used that were not as accurate as the UUT, meaning the calibration exercise could be “de-calibrating” the UUT. The example cited above was a real sample of the type of Standard Train Accuracy that was used to calibrate temperature indicators with a calibration limit of ±0.5oC. A Standard Train with an accuracy of ±1.03oC cannot be used to calibrate to ±0.5oC.
Lesson 2: There was an assumption that qualified techs would automatically check for an appropriate TAR, especially since an adequate TAR was identified as a requirement in the SOP. In fact, the techs did not have ready access to Standard accuracy information and were not evaluation TAR. The SOP statement was just that - a statement – not a direction to do anything. Techs must be explicitly directed to evaluate the TAR and document that evaluation.
As a result of these preliminary findings, several weeks of long were spent evaluation reverse traceability reports and calibration results and generating product investigations. All product shipments were held and a recall was considered but postponed until the results of the calibration assessments were completed. Remediation plans and approaches worked through, a protocol was designed to guide the assessment, Quality auditing of the process incorporated, rigid deadlines for completing the evaluation set, innumerable meetings with site and corporate management (at the highest level) were held.
Some general observations resulted from these assessments that are worth considering.
First, the database generated from the reverse traceability report showed inconsistencies in setting calibration limits for similar instruments/applications.
Second, the review showed that some limits were completely unreasonable. Differential pressure calibration limits of 0.0001“ H2O, for example. These limits make a TAR of greater than 1:1 impossible, let alone 4:1, using standard industrial Standards.
At this point, results of this exercise have pointed to the need to:
Lesson 3: Consolidate the Standards and reduce the number of models of Standards where possible to improve control of the Standards.
Lesson 4: Insure the techs have a method of readily determining the TAR before a calibration is begun.
Lesson 5: Require the techs to evaluate the TAR and document that evaluation before beginning a calibration.
This exercise also pointed out that the CCMS solutions currently available do not have the innate capability to do a TAR evaluation, and even if they did, it should not occur after a calibration is complete, but should be documented on the calibration sheet as generated. This would require assigning specific Standard Trains to a particular calibration.
I hope this narrative provides insight into a major pitfall and shows ways to avoid the same issues.