This is a ready-to-use worksheet for documenting the scientific justification behind a bracketed or matrixed stability design under ICH Q1D. It is the supporting evidence a reviewer or auditor reads when a reduced design is questioned; keep it under change control alongside the stability protocol it supports. Replace every <<FILL: ...>> placeholder with your own specifics. A worked specimen with real numbers follows the blank template. Confirm the current wording of ICH Q1D before you rely on any principle stated here.
Document control header
| Field | Entry |
|---|---|
| Matrix title | Bracketing and Matrixing Design Rationale, <<FILL: PRODUCT>> |
| Document number | <<FILL: MTX-ID>> |
| Version | <<FILL: version>> |
| Effective date | <<FILL: date>> |
| Linked stability protocol | <<FILL: protocol number>> |
| Owner | <<FILL: formulation / analytical development lead>> |
1. Purpose
Document, factor by factor, why a bracketed or matrixed design is scientifically justified for <<FILL: PRODUCT>>, what evidence supports treating the reduced-out levels as behaving like the tested ones, and which factors or time points were deliberately excluded from any reduction because they carry the most risk.
2. Full design and reduction summary
| Field | Entry |
|---|---|
| Full-design pull count (no reduction) | <<FILL: strengths x pack sizes x conditions x time points x batches>> |
| Reduced-design pull count | <<FILL>> |
| Percent reduction | <<FILL>> |
| Design type used | Bracketing only / Matrixing only / Both |
3. Bracketing decision, by factor
One row per factor considered for bracketing (strength, container size, fill volume, or similar). Do not omit a factor because bracketing it seemed obviously fine; the negative decision needs the same evidence trail as a positive one.
| Factor | Levels in the full design | Levels selected for testing (bracket extremes) | Levels covered by bracket, not tested | Similarity evidence | Bracketed? (Y/N) |
|---|---|---|---|---|---|
<<FILL: e.g. strength>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL: e.g. dose-proportional formulation, same degradation pathway in development data>> | <<FILL>> |
<<FILL: e.g. pack size / fill volume>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> |
4. Matrixing decision, on time points
| Field | Entry |
|---|---|
| Matrixing fraction | <<FILL: e.g. one-half reduction>> |
| Time points always tested in full, every combination (must include time zero and the final/longest time point) | <<FILL>> |
| Time points split across matrix cells | <<FILL>> |
| Number of matrix cells and the rotation plan | <<FILL>> |
| Balance check: is every combination and condition represented a comparable number of times across the design? | Yes / No, with correction if No |
| At least one combination has an unbroken, full-frequency profile across all time points? | Yes / No |
5. Factors and time points explicitly excluded from reduction
State what was deliberately kept at full frequency because development knowledge, risk assessment, or prior experience flagged it, and why.
| Factor / time point kept at full frequency | Reason it was excluded from bracketing or matrixing |
|---|---|
<<FILL>> | <<FILL: e.g. this strength used a different granulation step in development>> |
<<FILL>> | <<FILL: e.g. this is the significant-change risk window identified in forced degradation>> |
6. Risk assessment
| Question | Answer |
|---|---|
| If a bracketed-out level later shows a problem, is there any data to characterize it? | <<FILL: state the residual risk and mitigation, e.g. ongoing stability includes the bracketed-out strength every year to catch a late divergence>> |
| Is the reduced design applied to a high-risk product (narrow therapeutic index, complex delivery, biologic)? | <<FILL>> |
| If high-risk, is the design still full or minimally reduced, per the conservative-design principle? | <<FILL>> |
7. Acceptance criteria
- Every factor considered for bracketing has a documented similarity basis, not an assumption.
- The matrix design is balanced and includes at least one unbroken full-frequency profile.
- Time zero and the final/longest time point are tested for every combination, never matrixed away.
- No factor or time point identified as elevated risk in development or forced-degradation data was reduced without a separate, explicit justification.
- The reduction is consistent with the design actually written into the linked stability protocol.
8. References
ICH Q1D, Bracketing and Matrixing Designs for Stability Testing of Drug Substances and Drug Products. ICH Q1A(R2), Stability Testing of New Drug Substances and Products.
Confirm the current version and clause numbers before issue.
9. Revision history
| Version | Date | Author | Summary of change |
|---|---|---|---|
<<FILL: 1.0>> | <<FILL: date>> | <<FILL>> | Initial worksheet. |
10. Approvals
| Role | Name | Signature | Date |
|---|---|---|---|
| Formulation / analytical development | <<FILL>> | ||
| Quality Assurance | <<FILL>> | ||
| Regulatory affairs | <<FILL>> |
Filled specimen
A tablet marketed in three strengths (50 mg, 100 mg, 150 mg) in two pack sizes (30 count, 90 count), long-term and accelerated conditions, time points 0, 3, 6, 9, 12, 18, 24 months.
Section 2. Full design: 3 strengths x 2 pack sizes x 2 conditions x 7 time points x 3 batches = 252 pulls. Reduced design (strength bracket plus one-half time matrixing): approximately 110 pulls, a 56 percent reduction.
Section 3, bracketing. Strength: full-design levels 50/100/150 mg; tested levels 50 mg and 150 mg (extremes); 100 mg covered by bracket, not separately tested. Similarity evidence: all three strengths are dose-proportional dilutions of the same blend, sharing identical excipient ratios, and development forced-degradation data showed the same degradation products at the same relative rate across all three strengths. Bracketed: Yes. Pack size: full-design levels 30 count and 90 count; both tested (only two levels exist, so bracketing does not apply here; both are kept). Bracketed: No.
Section 4, matrixing. One-half reduction on time. Time zero and 24 months tested in full for every strength/pack-size combination. The 3, 6, 9, 12, and 18 month points are split across two rotating cells so every combination is tested at every time point in at least one cell. Balance check: Yes, each combination appears a comparable number of times across the two cells. Unbroken full profile: Yes, the 150 mg / 90 count combination is tested at every time point in both cells as the anchor profile.
Section 5, exclusions from reduction. The accelerated condition is never matrixed; it is tested in full for every combination and time point, since it decides whether the intermediate condition triggers. The 150 mg strength is kept in the design (not dropped, only reduced from a bracket to a tested extreme) because early development batches at this strength showed marginally faster degradant formation than 50 mg and 100 mg, even though within acceptable limits.
Section 6, risk assessment. If 100 mg later diverges, ongoing (annual) stability places at least one 100 mg batch on full long-term testing every year specifically to catch that divergence, since 100 mg carries no dedicated data in the registration design. Product is not high-risk (standard oral solid, wide therapeutic margin), so the reduction described here is considered proportionate.
Common inspection findings this matrix prevents
- A bracketed or matrixed design in the protocol with no rationale worksheet behind it, so a reviewer cannot tell whether the reduction was scientifically justified or simply convenient.
- Time zero or the final time point missing from full-frequency testing for some combinations.
- The strength or pack size most likely to diverge, based on development data, reduced away with no separate justification.
- A matrix design so unbalanced that some combinations are tested far more often than others with no stated reason.
- No plan to catch a bracketed-out level’s divergence during ongoing commercial-stage stability.
How to adapt this matrix
- List every factor actually varying across your product family, strength, pack size, fill volume, and decide bracketing and matrixing factor by factor, not as a blanket design choice.
- Pull the similarity evidence from real development or comparability data, not from an assumption that “they should behave the same.”
- Always keep time zero and the final/longest time point at full frequency across every combination.
- State explicitly which factor or time point was kept at full frequency because it carries the most risk, and why.
- Confirm every regulation in section 8 against the current published version before issue.