DOE for Analytical Method Validation (ICH Q2)

Part of our complete guide to DOE for biologists.

ICH Q2 lays out the characteristics an analytical method needs to demonstrate before you can trust it: specificity, linearity, range, accuracy, precision, and — the one DOE fits most naturally — robustness. Robustness asks a specific question: if a method parameter drifts a little from its target during routine use, does the result still hold up? That's a DOE screening problem almost by definition, and it's usually treated as an OFAT one instead.

Why one-at-a-time robustness testing under-tests the method

A typical OFAT robustness study nudges pH slightly, reruns the method, nudges temperature slightly, reruns it, and so on — one parameter at a time, everything else held at nominal. That tells you each parameter's individual sensitivity. It doesn't tell you what happens if pH and temperature both drift in the same direction during the same run, which is exactly the kind of combined, realistic deviation a method actually needs to survive in routine use. See DOE vs. OFAT for the general version of this gap — here it shows up specifically as an interaction between method parameters that OFAT robustness testing structurally can't see.

The DOE version of a robustness study

Take the method parameters you'd normally vary one at a time — pH, column temperature, flow rate, reagent concentration, whatever applies to your method — and set each at a small deliberate deviation from nominal (a low and high setting close to your target, not an extreme stress condition). A screening design run across all of them together, in a comparable or smaller number of runs than the OFAT version, tells you which parameters the method is actually sensitive to, including any that only matter in combination with another.

This is exactly the "Find what matters" screening goal in BioDOE — no different in kind from screening candidate CPPs in a process, just applied to method parameters instead. If a parameter turns out to matter more than expected, that's useful information before you finalize the method's operating tolerances, not after a batch fails downstream.

What this does and doesn't establish

A DOE robustness screen tells you which method parameters matter and, if you follow it with a characterization design, how sensitive the result is to each one across your intended range. It doesn't replace the other ICH Q2 characteristics — specificity, linearity, accuracy, and precision are separate validation exercises with their own designs, most of which aren't DOE problems in the same way. And as with any validation work, writing the protocol, setting acceptance criteria, and signing off on the validation report is the analyst's and QA's responsibility — BioDOE runs the experimental design and statistical analysis behind the robustness portion, not the validation itself.

Screening designs — the ones a robustness study needs — are available on the free Starter plan.

Start free and screen your method's parameters before you commit to tolerances.