What ICH E6(R3) actually changes for clinical sites
ICH E6(R3) is the biggest GCP update since 2016. The headlines are familiar — quality by design, risk-based monitoring, electronic systems — but the practical impact at sites looks different than the press releases suggest.
Read more →The R3 revision dropped in 2024 and replaced the R2 integrated addendum from 2016. Most coverage focused on the structural changes — the document is now organized differently with clearer principles vs. annexes — but the operational impact lands at three specific places in site work.
1. Quality by Design becomes a documented requirement
Under R2, "quality by design" was an aspirational concept. Under R3, sites are expected to document how their procedures are designed to prevent the most likely errors before they happen, not just detect them after. For most sites this means revising your site SOP for protocol training to include an explicit risk discussion at the SIV — not just a topic checklist.
2. Risk-based monitoring is now expected, not optional
Sponsors operating under E6(R3) are expected to implement risk-based monitoring. This pushes pressure down to sites: full SDV across every data point is no longer the default. Sites need to be ready for monitoring visits that focus on critical data and processes rather than line-by-line review.
3. Electronic systems get sharper requirements
R3 tightens the language around electronic data — audit trails, validated systems, controlled access. If your site uses a paper-electronic hybrid (still common in academic settings), you'll need to be explicit about which is the source of truth for each data point.
What to do today
- Review your site Quality Manual against the R3 quality-by-design language and update if needed
- Update protocol training documentation to capture risk discussion explicitly (not just topic checklists)
- Audit your hybrid paper-electronic workflows and document which is the source of truth per data element
- If you use validated electronic systems, confirm audit trail completeness against R3 expectations