ICH E8(R1): The Critical-to-Quality Factors ICH E6(R3) Assumes You Already Named

ICH E8(R1) reached Step 4 on October 6, 2021, and FDA published it as final guidance on April 11, 2022 — the first substantial revision of E8 since 1997. It does not create a new submission or a new form. It reframes what clinical trial quality is for: not a compliance record built to survive an audit, but a small set of critical-to-quality factors, named before the study starts, whose failure would actually matter. Most teams read E8(R1) once, nod at “quality by design,” and move on to E6(R3) — which is exactly backwards, because E6(R3)'s quality system has nothing concrete to manage until E8(R1)'s question has been answered.

What E8(R1) actually changed

The 1997 version of E8 was a descriptive overview of clinical study types. E8(R1) is a working document: it applies quality-by-design thinking to the study itself, addresses a broader range of study designs and data sources than the original, and updates cross-referencing to the rest of the ICH efficacy guidelines — most consequentially to ICH E6(R3)'s risk-based quality system. The shift is from quality as something inspected in afterward to quality as something designed in from the start.

What counts as a critical-to-quality factor

E8(R1) ties the definition to four questions: would this factor's failure compromise the protection of study participants, the integrity of the data, the reliability of the results, or the study's ability to meet its objectives? A factor that fails none of those tests is not critical to quality — it may still be worth doing well, but it does not belong on the short list. That distinction is the part teams skip.

• The primary endpoint's measurement. How it is captured, by whom, and what would make that measurement unreliable — not the endpoint's definition, which is already in the protocol.
• Eligibility criteria that drive interpretability. The criteria where a drift in enforcement would change what the result actually means, not every inclusion/exclusion line.
• Blinding integrity, where blinding matters to the result. The specific points where unblinding risk is real, not a generic statement that the study is blinded.
• Whatever the estimand you pick depends on. If the estimand assumes a particular handling of intercurrent events, the data that supports that handling is very likely critical to quality.

A risk register with forty items and a critical-to-quality list are not the same document. One is comprehensive. The other is supposed to be short enough that everyone on the study team can name it from memory. Why the CtQ list has to stay small

Why E6(R3) needs this list to already exist

ICH E6(R3)'s quality management system is built around risk-based monitoring: identify what matters, monitor that, and stop treating every site visit and every data point as equally worth the same scrutiny. That only works if the study already knows what matters. A sponsor that arrives at E6(R3) implementation without having done the E8(R1) work ends up building a risk-based monitoring plan around a generic risk matrix inherited from the last study, then calling it quality by design because the document has the right section headings. Clinical quality assurance teams asked to audit against E6(R3) frequently find this gap first: the monitoring plan references risk without any study-specific CtQ factors underneath it.

None of this requires a new department. It requires deciding, while the protocol itself is still being written, which few things would actually break the study if they went wrong — and writing that list down where the monitoring plan, the vendor oversight plan, and the data review plan can all reference the same short answer. Sponsors who treat E8(R1) as background reading end up building an E6(R3)-shaped quality system with no study-specific content inside it; the ones who do the naming work first get a quality system that actually protects something.