This is a ready-to-use critical-to-quality (CtQ) factor register for a clinical study, built on the ICH E8(R1) quality-by-design approach. It turns the abstract idea of “build quality in” into an operational tool: the factors that most affect participant safety and result reliability, each with a risk rating, a control, and a numeric tolerance limit that tells monitoring when to escalate. Replace every <<FILL: ...>> placeholder. A filled specimen follows. This is educational reference content, not regulatory advice.
Why this register earns its place
E8(R1) asks sponsors to identify the factors critical to a study’s quality and to focus effort there rather than verifying every field uniformly. The register is where that decision is recorded and made auditable. The tolerance limit column is what separates a real operational tool from a wish list: it tells the central and on-site monitoring teams exactly what to watch and when to act. See ICH E8(R1) and E9 trial design and statistics and risk-based monitoring in clinical trials.
Header
| Field | Entry |
|---|---|
| Register number | <<FILL: DOC-ID>> |
| Protocol number / title | <<FILL>> |
| Estimand reference | <<FILL: link to the primary estimand>> |
| Version / date | <<FILL>> |
| Cross-functional owner | <<FILL>> |
Prerequisite
Define the primary estimand first. The estimand tells you which data are critical to the primary analysis, which is the input the CtQ exercise needs. See the estimand specification template. Building the register before the estimand leads to guessing.
Column definitions
| Column | Meaning |
|---|---|
| CtQ factor | The study attribute assessed |
| Category | Scientific / Operational / Ethical |
| Why critical | Impact on participant safety or on reliability/interpretability of results |
| Risk rating | High / Medium / Low, from likelihood x impact x detectability |
| Control / design feature | How it is designed in and overseen |
| Tolerance limit | The quantitative or explicit threshold that triggers action |
| Monitoring focus | Where the effort goes (SDV, central/statistical monitoring, review) |
Register
| CtQ factor | Category | Why critical | Risk rating | Control / design feature | Tolerance limit | Monitoring focus |
|---|---|---|---|---|---|---|
<<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> |
Sign-off (cross-functional)
| Role | Name | Signature | Date |
|---|---|---|---|
| Clinical lead | <<FILL>> | ||
| Biostatistician | <<FILL>> | ||
| Data management | <<FILL>> | ||
| Clinical QA | <<FILL>> |
Review triggers
Revisit the register when an assumption breaks: slow enrollment, an endpoint behaving unexpectedly, a vendor change, or a protocol amendment. QbD is not a one-time gate.
Filled specimen
Excerpt for a phase 3 trial with a treatment-policy primary estimand on a week-24 endpoint.
| CtQ factor | Category | Why critical | Risk | Control / design feature | Tolerance limit | Monitoring focus |
|---|---|---|---|---|---|---|
| Primary endpoint measurement (6-minute walk test) | Scientific | Drives the primary analysis and the claim | High | Standardized SOP, central rater training and certification, equipment calibration check | Untrained rater performing any assessment: 0 tolerated | 100% source review of primary endpoint |
| Randomization integrity (IRT/IWRS) | Operational | Bias and unblinding destroy interpretability | High | Validated IRT, blinded kit labels, emergency unblinding log reviewed | Any unauthorized unblinding investigated as critical deviation | Central review of unblinding log |
| Eligibility: confirmed diagnosis at screening | Scientific | Wrong population invalidates inference | High | Independent central read of diagnostic imaging before randomization | 0 randomized without confirmed eligibility | 100% eligibility verification |
| Primary-endpoint completeness at week 24 | Scientific | Missing data threatens the estimand (post-discontinuation data still required under treatment policy) | High | Visit-window reminders, retention plan, central tracking of missing visits | Missing primary-endpoint rate above 8% triggers escalation | Central statistical monitoring of missing-data rate |
| Concomitant medication coding (non-rescue) | Operational | Supportive, not key to primary analysis | Low | Standard coding dictionary, periodic review | Routine query handling | Sampled review only |
The specimen shows the estimand driving the register: because the primary estimand uses a treatment-policy strategy, week-24 data must be chased even after discontinuation, so completeness at week 24 becomes a High CtQ factor with a numeric tolerance (8%). A register that rated it Low would have quietly broken the estimand.
Common inspection findings this register prevents
- CtQ factors never defined, so monitoring is undifferentiated: everything verified at huge cost, or nothing prioritized and critical data unchecked.
- Tolerance limits written as unauditable words (“low missing data”) instead of numbers.
- The register built without the estimand, so the data the statistics demand are not flagged as critical.
- No cross-functional sign-off, so QA and data management never operationalized the consequences.
How to adapt this register
- Define and link the primary estimand first; let it tell you which data are critical.
- Write tolerance limits as numbers wherever the factor allows.
- Point the monitoring-focus column at your real monitoring plan and central-statistical-monitoring setup.
- Get cross-functional sign-off, including QA and statistics, and set review triggers.