In a series of posts, we are going to talk about method validation.
- Part 1: Introduction-Is it valid, invalid or non-validated?
- Part 2: What is method validation?
- Part 3: Can we use someone else’s validated method?
- Part 4: What triggers verification, re-validation or out right new validation of a method?
- Part 5: What are the essential terms in method validation?
- Part 6: What is quality assurance and quality control?
From time to time all instruments have error. This is just a fact of analytical testing. Machines break down, maintenance is needed, changes in the configuration of the machines, software upgrades and the like are not infrequent occurrences. The question becomes when do we need to start from scratch and when do we simply need to verify the validation.
There are events that trigger verification of the method, re-validation of the method, and even completely new method validation studies.
According to Huber, analytical methods need to be validated or re-validated
- before their introduction into routine use;
- whenever the conditions change for which the method has been validated (e.g., an instrument with different characteristics or samples with a different matrix); and
- whenever the method is changed and the change is outside the original scope of the method.
A real life example of this would be when we switch from one operating system to another. We will need to conduct a full scale validation study from scratch. If we change the chromatographic conditions of the oven, for example, in GC-FID work, then we would need to validate the entire method anew. If we switch from sampling and testing urine to that of whole blood, we need to validate from scratch. If we change a consumable such as a column or an inlet liner, we need to re-validate (robust verification) before its employed sampling the unknowns again. If we re-calibrate our pipettes, then re-validation (robust verification) is required. If we do something minor, such as change the in the report that is generated that the acquisition number follows the subject’s name instead of vice versa, then we do not need to validate anew or re-validate at all. In essence, when in doubt, validate anew.
The data from the method validation should be “clean” and it should represent the most perfect application of the method possible in that particular environment on that particular instrumentation under those set conditions.
As a logical result, if we see errors in the data in the validation efforts, then we cannot trust that when unknowns are sampled and tested later that the results will be error free.
We have to really scrutinize the underlying data in the validation study to determine fairly and objectively whether or not the data supports the contention that the method is indeed valid, meaning that it is suitable for its intended purpose or whether it is just simply someone’s opinion with no data to support it.