This is a ready-to-use checklist for the step that most product reviews skip past under schedule pressure: confirming the data pull is actually complete and reconciled before anyone starts trending it. A review built on an unreconciled data set produces trending that looks rigorous and means nothing. Complete this checklist once per review cycle, before section 5.5 (trending) of the governing SOP begins. Mark each item Pass, Fail, or NA, record the evidence reference, and resolve any Fail before trending starts. Replace every <<FILL: ...>> placeholder. A filled specimen follows.
How to use this checklist
- Complete one checklist per product per review cycle, before trending begins.
- Every item marked NA needs a one-line reason.
- Any Fail blocks the start of trending until resolved; do not trend an unreconciled data set “provisionally.”
- The person completing this checklist should not be the sole source of the underlying data extracts, so the reconciliation is an independent check, not a self-check.
Header
| Field | Entry |
|---|---|
| Product / strength / presentation(s) covered | <<FILL>> |
| Review period | <<FILL: from date>> to <<FILL: to date>> |
| Data cutoff date | <<FILL>> |
| Reviewer | <<FILL>> |
| Review date | <<FILL>> |
| Linked APR/PQR report reference | <<FILL>> |
Section A: Register counts reconcile
| # | Item | Pass / Fail / NA | Evidence reference | Basis |
|---|---|---|---|---|
| A1 | Batch disposition count in the data pull matches the production schedule / disposition system for the exact period | <<FILL>> | Governing SOP section 5.4 | |
| A2 | OOS count in the data pull matches the laboratory OOS log for the exact period, batch for batch | <<FILL>> | 21 CFR 211.192 | |
| A3 | Deviation count in the data pull matches the deviation log for the exact period | <<FILL>> | 21 CFR 211.192 | |
| A4 | Complaint count in the data pull matches the complaint-handling system for the exact period | <<FILL>> | 21 CFR 211.198 | |
| A5 | Stability data pulled matches the stability program’s schedule of annual commitment batches and active studies covering the period | <<FILL>> | ICH Q1A(R2) | |
| A6 | Change control count in the data pull matches the change control system for the exact period | <<FILL>> | Change control procedure | |
| A7 | Returns / salvage / recall data matches the respective logs for the exact period | <<FILL>> | GMP |
Section B: Batch list ties to disposition
| # | Item | Pass / Fail / NA | Evidence reference | Basis |
|---|---|---|---|---|
| B1 | Batches manufactured equals batches released plus rejected plus reworked plus on hold, with no orphan batches | <<FILL>> | Governing SOP section 5.4 | |
| B2 | Every batch on the list carries a disposition status current as of the data cutoff | <<FILL>> | ||
| B3 | Batches still open (in process, on hold, or under investigation) at cutoff are listed as open, not omitted | <<FILL>> | ||
| B4 | Rejected and investigated batches are included in the quantitative trending set, not excluded as “unrepresentative” | <<FILL>> | 211.180(e), “representative batches” |
Section C: Data granularity
| # | Item | Pass / Fail / NA | Evidence reference | Basis |
|---|---|---|---|---|
| C1 | Data arrives at batch level or individual result level, not as a pre-digested narrative summary (“all conforming”) | <<FILL>> | Governing SOP section 5.3 | |
| C2 | Where a contract manufacturing organization or contract lab contributed data, it arrived at the level of detail the quality agreement requires, sufficient to trend independently | <<FILL>> | Quality agreement reference | |
| C3 | Categorical data (deviations, OOS, complaints) is coded by type/category in the extract, not left as free text requiring manual re-categorization | <<FILL>> |
Section D: Exclusions and cutoff discipline
| # | Item | Pass / Fail / NA | Evidence reference | Basis |
|---|---|---|---|---|
| D1 | Every record excluded from quantitative trending has a written reason on file | <<FILL>> | ||
| D2 | No batch, deviation, OOS, or complaint was excluded solely because it was unfavorable | <<FILL>> | ||
| D3 | The data cutoff date was honored consistently across every source system queried (no system queried on a different date without justification) | <<FILL>> | ||
| D4 | Every extract carries its source system, query identity or report ID, and extraction date | <<FILL>> | Re-derivability | |
| D5 | The reviewer completing this checklist is independent of the sole extraction of the data (not the only person who could reproduce the query) | <<FILL>> |
Signoff
| Field | Entry |
|---|---|
| All items Pass or NA (with reason), or Fails resolved | Yes / No |
| Open items blocking start of trending | <<FILL: none, or list>> |
| Reviewer (name, signature, date) | <<FILL>> |
| Cleared to begin trending on | <<FILL: date>> |
Acceptance criteria
The data pull is ready for trending when every item above is Pass or a justified NA, every register count reconciles, the batch list ties to disposition with no orphans, all data arrived at batch or result level, every exclusion is documented, and the cutoff date was honored consistently across every source queried.
Filled specimen (extract)
The following shows a completed extract for an illustrative solid oral dose product’s 2026 annual cycle. The product and numbers are illustrative.
| # | Item | Pass / Fail / NA | Evidence reference |
|---|---|---|---|
| A2 | OOS count matches lab log | Pass | 7 OOS in LIMS query LQ-2027-018 = 7 OOS in lab OOS log for the period |
| B1 | Batches manufactured = released + rejected + reworked + on hold | Pass | 42 = 39 + 1 + 1 + 1, PROD-2027-002 |
| B4 | Rejected and investigated batches included in trending | Pass | Batch FT-10-027 (rejected, OOS-2026-0091) included in section 4 trending, not excluded |
| C2 | CMO data at trendable detail | Fail, resolved | Site C’s initial submission was a one-line summary; batch-level resubmission requested and received 22 Jan 2027, now Pass, reference CMO-DATA-2027-006 |
| D3 | Cutoff honored consistently | Pass | All 6 source systems queried against the 31 Dec 2026 cutoff, query log QL-2027-011 |
The reviewer here caught that the CMO’s first submission (item C2) was a non-trendable summary, rejected it, and required the batch-level resubmission before clearing the data set for trending. That is exactly the check this checklist exists to force before anyone builds a control chart on top of data that cannot actually be trended.
Common inspection findings this checklist prevents
- Trending built on a data set that was never reconciled against the source systems, discovered only when an inspector’s own count of OOS results does not match the report.
- A batch list with orphan batches, batches manufactured but not accounted for in any disposition category.
- A CMO or contract lab’s “all conforming” summary accepted as the data pull instead of trendable batch-level data.
- Exclusions from trending with no documented reason, indistinguishable from cherry-picking.
- Different source systems queried against different cutoff dates without anyone noticing, so the review’s own internal counts do not agree with each other.
How to adapt this checklist
- Add or remove rows in Section A to match your actual data source inventory (add rows for equipment/utility qualification status and MA-variation tracking if those live in separate systems from the ones listed).
- Point the “Basis” column at your own SOP numbers alongside the regulation citations.
- Set your rule for who may complete this checklist versus who authored the underlying extracts, preserving the independence in item D5.
- Confirm each regulation reference against the current published version before issue.