A Validation Sampling Strategy describes what will be sampled, where samples will come from, how many samples will be taken, when sampling will occur and how the data collected will provide evidence of process control. In pharmaceutical validation, sampling is more than a laboratory testing process – it is part of the scientific evidence for the successful completion of a process.
Poor sampling plans can negatively impact the validity of an otherwise well-designed validation study. If the samples are taken at convenient locations, at random times and with a restricted number of samples, it will lead to poor results.
The FDA process validation guidance highlights sampling places, the sampling size, the sampling frequency and the link between sampling and statistical confidence as significant metrics in the process qualification protocol.
Rather, a more sensible question ought to be:
What process or quality attribute is intended to be demonstrated and what type of sampling would prove to deliver insightful data?
As a result, the sampling strategy ought to be geared toward the corresponding critical quality attributes (CQAs), critical process parameters (CPPs), process risks and the precise aim of validation study.
To provide an example, the sampling issue faced by the blend uniformity experiment would differ from that faced by the cleaning validation study. The validation process of tablet weight can require sampling throughout the production process, while the validation process of cleaning would rather focus on hard-to-clean places.
For example, in terms of validation of the processes within a plant, one must consider:
According to the FDA guidelines, the samples collected should represent the batch of material in question and must guarantee the necessary statistical sufficiency when the samples are taken or tested in order to control the performance of the process.
This is especially important in relation to cleaning validation, equipment qualification, environmental monitoring and process validation. In each of these instances, there can be varying levels of risk associated with different physical locations.
For cleaning validation the difficult areas might include dead ends, gaskets, valves, tight spaces, equipment joints, shadowing by spray pattern, as well as places which are difficult to access. The use of samples taken only from easily accessible stainless steel surfaces may produce a distorted image of cleaning efficacy.
In the process validation area selection of locations is determined by the process design. For instance, in a blending operation it would make sense to look at different locations in the blender instead of relying on the same discharge sample.
From the QA standpoint I would expect the protocol explaining the choice of important sampling locations rather than just listing them.
The amount of samples must be justified based on scientific reasons, including aspects like variability of the process, risk, size of the batch, complexity of the process, desired confidence, data available and the feature in question.
FDA advises to have a sufficient number of samples during PPQ to provide a desired level of statistical confidence about the quality within and between batches. This margin of confidence may depend on risk analysis of the variable being measured.
Thus, saying "three samples are taken because three samples is a standard number" is not enough unless it can be explained why three samples was the optimal number of samples for the study.
Whereas a statistically designed sample size may be justified for some features, some other variables may require a risk-based scientific reasoning to justify the sample size.
It is essential to notice that the number of samples should be justified before the beginning of a study rather than after looking at the results of the study.
For example, a key parameter may vary during a long production cycle and collecting just the first and last samples may skip the fluctuations that occur in the middle.
According to the FDA, PPQ requires more sampling, testing and inspection than regular commercial manufacturing. The additional control should help to determine proper regular sampling levels during the process verification period.
That means that the sampling plan needs to take into account possible variations:
When determining worst-case scenarios, a batch, item, equipment, or location should not just be selected because it is considered to be the hard, however, the reasons for such determination must be supported through the assessment of risks and process knowledge.
In case of cleaning validation, no cleaning product should be considered worst-case unless its solubility, potency, toxicity, cleanability and the role of the product in relation to cleaning equipment have been analyzed. In the process validation, worst-case operational conditions should involve difficulty but provide acceptable range of critical process parameters or characteristics of the material used.
If a document states that a location or a situation is considered worst case, an inspector might inquire as follows:
What are the reasons for this claim?
The reasons for that must be supported by development data, as well as risk assessment and other objective evidence.
Another thing that can lead to trouble is changing the sampling strategy after results of testing differ from expectations. When new samples are taken due to initial testing outcomes, detailed analysis of why it was needed and how testing results are affected must take place.
Validated plan defines sampling procedure and acceptance criteria. Whenever any deviations from originally agreed procedure take place, such deviations have to be documented, evaluated and their effects on the outcomes have to be analyzed.
Records of sampling need to contain information about samples’ identity, time of collection, batch/lot association, sampling interpretations, as well as samples’ handling details and data integrity.
An ideal technique integrates together the risk assessment process, knowledge about the process itself, as well as the location and quantity of sampled material, sampling frequency, analytical techniques and checking criteria.
In my reviews of validation plans I look for such connection since if the sampling plan is unable to explain which variability it aims to catch and why it has chosen such particular sample, the validation plan is not complete.
Thus, an efficient validation sampling plan is more than produce test outcomes. It makes sure the data gathered represent the validated process, thus providing reliable proof of its control state.
The FDA process validation guidance highlights sampling places, the sampling size, the sampling frequency and the link between sampling and statistical confidence as significant metrics in the process qualification protocol.
Start With the Attribute Being Demonstrated
The most sensible question to ask should not be "How many samples do we need?"Rather, a more sensible question ought to be:
What process or quality attribute is intended to be demonstrated and what type of sampling would prove to deliver insightful data?
As a result, the sampling strategy ought to be geared toward the corresponding critical quality attributes (CQAs), critical process parameters (CPPs), process risks and the precise aim of validation study.
To provide an example, the sampling issue faced by the blend uniformity experiment would differ from that faced by the cleaning validation study. The validation process of tablet weight can require sampling throughout the production process, while the validation process of cleaning would rather focus on hard-to-clean places.
Sampling Must Represent the Process
Before any sampling can be undertaken, I would always check to see if the samples which are to be taken are actually representative of the product under consideration.For example, in terms of validation of the processes within a plant, one must consider:
- Beginning, middle and end of the manufacturing run
- Locations of equipment used
- Stages of processing used
- Batches processed
- Shifts or other operational condition
- Worst case conditions used in the process
- Locations where products of a higher risk of variability exist scientifically
According to the FDA guidelines, the samples collected should represent the batch of material in question and must guarantee the necessary statistical sufficiency when the samples are taken or tested in order to control the performance of the process.
Sample Location Should Come from Risk Assessment
The location from which the sample is taken must have an established scientific justification.This is especially important in relation to cleaning validation, equipment qualification, environmental monitoring and process validation. In each of these instances, there can be varying levels of risk associated with different physical locations.
For cleaning validation the difficult areas might include dead ends, gaskets, valves, tight spaces, equipment joints, shadowing by spray pattern, as well as places which are difficult to access. The use of samples taken only from easily accessible stainless steel surfaces may produce a distorted image of cleaning efficacy.
In the process validation area selection of locations is determined by the process design. For instance, in a blending operation it would make sense to look at different locations in the blender instead of relying on the same discharge sample.
From the QA standpoint I would expect the protocol explaining the choice of important sampling locations rather than just listing them.
How Many Samples are Enough?
An appropriate number of samples cannot be set once and applied across all validations.The amount of samples must be justified based on scientific reasons, including aspects like variability of the process, risk, size of the batch, complexity of the process, desired confidence, data available and the feature in question.
FDA advises to have a sufficient number of samples during PPQ to provide a desired level of statistical confidence about the quality within and between batches. This margin of confidence may depend on risk analysis of the variable being measured.
Thus, saying "three samples are taken because three samples is a standard number" is not enough unless it can be explained why three samples was the optimal number of samples for the study.
Whereas a statistically designed sample size may be justified for some features, some other variables may require a risk-based scientific reasoning to justify the sample size.
It is essential to notice that the number of samples should be justified before the beginning of a study rather than after looking at the results of the study.
Importance of the Sampling Frequency
The sampling frequency influences the ability of the study to monitor the process behavior over time.For example, a key parameter may vary during a long production cycle and collecting just the first and last samples may skip the fluctuations that occur in the middle.
According to the FDA, PPQ requires more sampling, testing and inspection than regular commercial manufacturing. The additional control should help to determine proper regular sampling levels during the process verification period.
That means that the sampling plan needs to take into account possible variations:
- At a certain stage of the manufacturing process
- After recalibration
- During arriving at a new production cycle
- In the course of variable conditions
- Throughout a long-term operation
- Between different batches
Worst-Case Sampling Needs Evidence
Hair-raising situations arise in worst-case sampling.When determining worst-case scenarios, a batch, item, equipment, or location should not just be selected because it is considered to be the hard, however, the reasons for such determination must be supported through the assessment of risks and process knowledge.
In case of cleaning validation, no cleaning product should be considered worst-case unless its solubility, potency, toxicity, cleanability and the role of the product in relation to cleaning equipment have been analyzed. In the process validation, worst-case operational conditions should involve difficulty but provide acceptable range of critical process parameters or characteristics of the material used.
If a document states that a location or a situation is considered worst case, an inspector might inquire as follows:
What are the reasons for this claim?
The reasons for that must be supported by development data, as well as risk assessment and other objective evidence.
A Common Validation Sampling Mistake
One of the things that often goes wrong in sampling is creating sampling plan focusing on ease of collecting samples instead of what is required for understanding the process.Another thing that can lead to trouble is changing the sampling strategy after results of testing differ from expectations. When new samples are taken due to initial testing outcomes, detailed analysis of why it was needed and how testing results are affected must take place.
Validated plan defines sampling procedure and acceptance criteria. Whenever any deviations from originally agreed procedure take place, such deviations have to be documented, evaluated and their effects on the outcomes have to be analyzed.
Records of sampling need to contain information about samples’ identity, time of collection, batch/lot association, sampling interpretations, as well as samples’ handling details and data integrity.
What QA should Review Before Approval
In granting approval of a validation protocol, I would ask a few pragmatic questions:- Is there a sampling scheme developed for any critical attribute?
- Are sampling sites supported with a rationale?
- Is the number of samples validated with a risk or statistical argument?
- Does the frequency account for variability in the process?
- Are the worst-case sites or conditions sufficiently justified?
- Are the sampling methods validated when necessary?
- Were acceptance criteria determined in advance?
- Are the resulting data able to prove that the validation was successfully achieved?
- Has the system of sample identification and data storages been put in order?
Sampling is Part of the Validation Argument
Validation sampling is not an isolated lab activity conducted merely for the sake of filling a report. Rather, it constitutes one part of a larger scientific case for the efficacy of a given process in producing an acceptable end-product.An ideal technique integrates together the risk assessment process, knowledge about the process itself, as well as the location and quantity of sampled material, sampling frequency, analytical techniques and checking criteria.
In my reviews of validation plans I look for such connection since if the sampling plan is unable to explain which variability it aims to catch and why it has chosen such particular sample, the validation plan is not complete.
Thus, an efficient validation sampling plan is more than produce test outcomes. It makes sure the data gathered represent the validated process, thus providing reliable proof of its control state.
Get ready to use editable Validation Protocols in MS-Word FormatView List
No comments:
Post a Comment
Please don't spam. Comments having links would not be published.