A method validation proves the test, not the batch

This blog's notes this week on what a cleaning validation protocol has to prove and what a stability protocol has to prove both assumed something neither one actually establishes: that the analytical test producing the numbers in the first place can be trusted. A cleaning swab result and a stability pull both resolve to a number from an assay, and the document that earns the right to trust that number isn't the validation protocol around the process — it's a separate validation, of the method itself, governed by its own guideline.

A process validation borrows a method's credibility; it doesn't create it

IQ, OQ, and PQ, covered earlier on this blog, qualify equipment and prove a process performs consistently. A stability or cleaning protocol proves a product or a surface behaves the way its specification predicts. Every one of those conclusions depends on an analytical procedure — an assay, a swab-and-recovery method, a dissolution test — reporting a number that reflects the sample and not the method's own noise. None of those protocols validates the method doing the measuring; they each assume it, and borrow its credibility without ever having established it themselves.

Q2(R2) is the current guideline, and it's a different document than Q2(R1)

ICH finalized Q2(R2), Validation of Analytical Procedures, at Step 4 on 1 November 2023, as a full rewrite rather than a minor update to the prior Q2(R1) text, built to align with the new ICH Q14 guideline on analytical procedure development. Its stated scope covers analytical procedures used for release and stability testing of commercial drug substances and drug products, with the guideline itself noting it can also be applied, using a risk-based approach, to other procedures that support a product's control strategy, and in a phase-appropriate way to procedures used earlier, in clinical development. A method validation protocol built against the superseded Q2(R1) text alone is citing a guideline ICH itself has already replaced for the product types it names.

The characteristics a protocol has to show, not just claim

What a method validation protocol actually has to demonstrate hasn't moved since the vocabulary ICH Q2 established: specificity, that the method measures what it claims to, undisturbed by related substances; accuracy; precision, split into repeatability and intermediate precision; linearity and range; detection and quantitation limits, where relevant to the test's purpose; and robustness, the method's tolerance for the small variations — a different column lot, a slightly different mobile-phase ratio — that happen in routine use. A protocol that states a bottom-line acceptance criterion without running each applicable characteristic as its own documented exercise hasn't validated the method; it's validated a single point on a curve it never drew.

Borrowing the release spec's criteria isn't validating the method

The shortcut that method validation shares with the stability protocols this blog described earlier this week is the same one: copying acceptance numbers from somewhere else — here, the release specification's pass-fail limit — and calling the copy a validation. A release specification states what result counts as acceptable for the product. It says nothing about how reproducible that result is between analysts, how the method behaves near its detection limit, or whether a structurally related impurity would register as the active ingredient. A method validation protocol that never runs those questions separately from the spec has produced a document shaped like a validation without the evidence a validation is supposed to contain.

Where this meets the file

A method validation protocol template that keeps each Q2(R2) characteristic — specificity, accuracy, precision, linearity, range, the relevant limits, robustness — as its own required, evidenced section, distinct from the release specification it eventually supports, is previewed in the launch catalog. If your lab structures that validation differently, the shelf takes that correction directly.