Analytical method validation (revision 4)
Old revision·15:40, 19 Aug 2024·KarlFischerKit
| Analytical method validation | |
|---|---|
| Governing guidance | ICH Q2(R2) |
| Core characteristics | Specificity, linearity, range, accuracy, precision, detection and quantitation limits |
| Establishes | Fitness for a stated purpose, not correctness in general |
| Analytical method infobox · conventions | |
Analytical method validation is the documented demonstration that an analytical procedure is suitable for its intended purpose. It is not a demonstration that a method is correct in the abstract: a method validated for assay of a drug substance is not thereby validated for detection of a trace impurity, and the purpose is part of the claim.[1]
The characteristics assessed are specificity, linearity, range, accuracy, precision (repeatability and intermediate precision), the detection and quantitation limits where relevant, and robustness. Which of these apply depends on what the method is for, and the compendial treatment of the same subject is framed the same way.[1][2]
The characteristics
[edit]| Characteristic | Question it answers |
|---|---|
| Specificity | Does the method measure the analyte and not something else? |
| Linearity | Is response proportional to concentration over the range? |
| Range | Between what limits is the method demonstrated to work? |
| Accuracy | Does the result agree with the true value? |
| Repeatability | Do replicate determinations agree within a run? |
| Intermediate precision | Do they agree across days, analysts and instruments? |
| Limit of detection | Below what level is presence undetectable? |
| Quantitation limit | Below what level is a number unreliable? |
| Robustness | Does the result survive small deliberate changes in conditions? |
Robustness is the characteristic most relevant to reading a third-party result. A method sensitive to a two-degree change in column temperature or a small change in mobile-phase pH will give different answers in different laboratories, and a difference between two results may reflect that rather than a difference in material.[1]