Key takeaway
A proposed 24-month shelf life does not turn month 18 or 24 into observed long-term data. Preserve the actual observation horizon and assess any extrapolation as a separate claim.
A duration field can quietly become the wrong claim
A pharmaceutical archive owner receives a summary with one field: “stability duration 24 months.” The source packet contains long-term results through month 12, accelerated results through month 6 and an assessment proposing a 24-month shelf life. The single field has merged observations under different conditions with a conclusion about a longer period.
The first owner decision is representational: can a reviewer tell which results were actually obtained, for which batches, formulation, container and attributes? Keep that question separate from whether a proposed shelf life is scientifically and regulatorily justified. Restoring the evidence boundary does not require inventing an expiry decision.
The pinned ICH Q1A(R2) framework distinguishes long-term and accelerated study conditions and ties product stability evidence to formulation and container-closure scope. Its scope is specific; this article does not extend those recommendations to every product, application type or jurisdiction.
Complete the observation matrix before reading the proposal
The hypothetical matrix invents batches B1, B2 and B3, formulation F1, container K1 and recorded attribute A. Assume the packet contains supported results at the listed times for each batch. The example does not state any acceptance criteria or claim that A alone is sufficient for a shelf-life assessment.
The matrix corrects the export: the observed long-term horizon is 12 months for the stated scope; the accelerated horizon is 6 months under its different conditions. The 24-month figure belongs to the separate assessment. A different container K2 is not silently populated with K1 results. Empty future cells mean “not observed in this packet,” not “failed” or “proven stable.”
| Scope and condition | Retained time points in the fixture | Corrected interpretation |
|---|---|---|
| B1–B3; F1/K1; attribute A; long-term | Months 0, 3, 6, 9 and 12 | Long-term observations through month 12 |
| B1–B3; F1/K1; attribute A; accelerated | Months 0, 3 and 6 | Accelerated observations through month 6 |
| B1–B3; F1/K1; long-term months 18 and 24 | No measurements in the packet | Not observed; do not manufacture results |
| F1/K1; proposed shelf life | 24 months in assessment E1 | Separate proposed conclusion; review its rationale |
| F1/K2; changed container | No matched study evidence supplied | Different scope; K1 observations do not fill its cells |
Allow an extrapolation without relabeling it as a measurement
FDA’s final Q1E guidance discusses extrapolation beyond the period covered by long-term data. That is a conditional inference, informed by knowledge of change patterns, the adequacy of the analysis and supporting data. Additional long-term observations are used to verify an extrapolated conclusion. The guidance is nonbinding; it does not supply an automatic conversion from accelerated months to approved shelf life.
For the fictional E1 assessment, preserve its author, stated product scope, attributes considered, data cutoff and reasoning beside the matrix. Link any continuing-study commitments separately. A qualified reviewer can then evaluate the proposal without losing the distinction between its supporting observations and the inference drawn from them.
This distinction works in both directions. It would be wrong to insert invented month 24 results because the proposal says 24 months. It would also be wrong to declare every extrapolation invalid merely because month 24 has not yet been observed. The actual scientific and regulatory assessment must address the applicable conditions and evidence.
Pin the edition and the status that were actually read
The EMA Q1A(R2) PDF used here has an August 2003 cover and a 2006 copyright notice. Those dates describe different things; the copyright notice does not turn it into a 2006 revision. FDA’s final Q1E PDF is dated June 2004. These month-and-year labels should not acquire invented publication days in an archive catalog.
The consolidated Q1 guidance listed by FDA in June 2025 remained marked Draft and Not for implementation on 16 October 2026. That status matters when identifying the basis of an older assessment or planning a new review. The draft listing is used here only as status evidence, not as an implemented replacement for the pinned final material.
Keep the jurisdiction, document edition and assessment date together. A current download address does not make a historical assessment conform to a later framework, nor establish that one jurisdiction has adopted another’s guidance.
Replace the ambiguous summary with two bounded statements
The completed owner summary now says: “For B1–B3/F1/K1 and the stated attributes, retained long-term observations extend through 12 months and accelerated observations through 6 months. Assessment E1 separately proposes 24 months; its extrapolation rationale and commitments require qualified review.”
This description permits a reviewer to request missing attributes or scope evidence without treating future observations as present. It is an evidence inventory, not product approval, a patient-use recommendation or an authorization to extend an actual expiry date.