This is a ready-to-use checklist for scoring how mature a Continued Process Verification (CPV) program actually is, against four stated levels, rather than assuming maturity because the program has existed for a while or produces a chart every batch. Run it on a defined cycle, not once as a training exercise, and keep every completed assessment on file so the program’s progress across cycles is itself a record. Replace every <<FILL: ...>> placeholder. A filled specimen follows.
Document control header
| Field | Entry |
|---|---|
| Parent CPV plan document number | <<FILL>> |
| Assessment type | <<FILL: scheduled / triggered>> |
| Trigger, if applicable | <<FILL: e.g. following a significant process change, a repeat inspection finding, a new product entering CPV>> |
| Assessor(s) | <<FILL>> |
| Assessment date | <<FILL>> |
Maturity levels
| Level | What it looks like |
|---|---|
| 1, Reactive | Charts exist and update, but review is a signature on a routing sheet; signals are explained verbally if at all, with little or nothing on record; parameter selection has never been traced back to the Stage 1 risk assessment |
| 2, Documented | Every parameter traces to Stage 1; escalation follows a written rule set; signals are logged, not just discussed; control limits are never set equal to specification; low-volume parameters use a stated provisional method rather than a silent thin chart |
| 3, Managed | The periodic trend summary and the annual product review consistently agree because the same people are drawing on the same charts; a post-approval process change triggers a parameter-specific CPV review, not a blanket reset or no reset at all; escalation timelines are tracked and met, not just written down |
| 4, Predictive | A multivariate model runs where correlated-parameter structure justifies it, with its own training set discipline and change control; the program runs this self-assessment on a defined cycle and closes the gaps it finds |
Checklist
| # | Item | Maturity dimension | Pass / Fail / N/A | Evidence / notes |
|---|---|---|---|---|
| 1 | Every monitored parameter traces to a stated Stage 1 source, and the traceability matrix is current | Level 2 | <<FILL>> | <<FILL>> |
| 2 | No control limit on any chart is set equal to, or wider than, the registered specification | Level 2 | <<FILL>> | <<FILL>> |
| 3 | Every signal from the stated period has a logged entry in the trend signal investigation and escalation log, including signals later explained as noise | Level 2 | <<FILL>> | <<FILL>> |
| 4 | Every low-volume or newly added parameter has a stated provisional method and a reconciliation trigger, not an undisclosed thin chart | Level 2 | <<FILL>> | <<FILL>> |
| 5 | Escalation timelines stated in the CPV signal response SOP were met this period, or a documented, approved extension is on file for every miss | Level 3 | <<FILL>> | <<FILL>> |
| 6 | The most recent periodic trend summary and the most recent annual product review cite the same open-signal status for the period they share | Level 3 | <<FILL>> | <<FILL>> |
| 7 | Every process change implemented through change control during the assessment period was reviewed for which specific monitored parameters it affects | Level 3 | <<FILL>> | <<FILL>> |
| 8 | Every control limit recalculation on file cites a change-control record dated before or concurrent with the data shift it explains | Level 3 | <<FILL>> | <<FILL>> |
| 9 | Correlated-parameter groups have been identified and evaluated for whether a multivariate model is warranted, even if the answer is not yet | Level 4 | <<FILL>> | <<FILL>> |
| 10 | Where a multivariate model is in use, it has its own written plan with training set qualification and retraining triggers on file | Level 4 | <<FILL>> | <<FILL>> |
| 11 | This self-assessment itself has been run on a defined, repeating cycle, with prior assessments on file showing closed gaps, not only the current one | Level 4 | <<FILL>> | <<FILL>> |
Scoring and maturity band
| Band | Basis |
|---|---|
| Level 1 | Items 1 through 4 show a majority of Fails |
| Level 2 | Items 1 through 4 Pass; a Fail exists among items 5 through 8 |
| Level 3 | Items 1 through 8 Pass; a Fail or N/A-without-rationale exists among items 9 through 11 |
| Level 4 | Items 1 through 11 Pass, or N/A with a documented, defensible rationale (for example, item 10 is N/A where item 9 correctly concludes no multivariate model is warranted yet) |
An N/A answer on items 9 or 10 is only acceptable with a stated reason; an unexplained N/A is scored as a Fail for banding purposes.
Disposition
| Field | Entry |
|---|---|
| Overall maturity level | <<FILL: 1 / 2 / 3 / 4>> |
| Gaps identified, with owner and target date | <<FILL>> |
| Next scheduled assessment date | <<FILL>> |
Approvals
| Role | Name | Signature | Date |
|---|---|---|---|
| Assessor (CPV program owner) | <<FILL>> | ||
| Approver (Quality Assurance) | <<FILL>> |
Filled specimen
Acme Therapeutics ran this assessment on its oligonucleotide drug substance CPV program on its standard annual cycle. Company and numbers are illustrative.
| # | Item | Result | Evidence / notes |
|---|---|---|---|
| 1 | Stage 1 traceability | Pass | Matrix current, last reconciled against the Stage 1 risk assessment three months prior |
| 5 | Escalation timelines | Pass | All three signals this period closed or escalated within the SOP’s stated windows |
| 6 | Periodic report vs. APR agreement | Pass | Q3 periodic report’s open item (resin lot investigation, CAPA-2026-0143) restated identically in the FY2026 APR draft |
| 9 | Correlated-parameter evaluation | Fail | No formal correlation analysis has been run across the program’s CQAs; several plausible candidates, full-length purity and related impurity levels, have never been checked |
| 10 | Multivariate plan | N/A, tied to item 9 | Not applicable until item 9’s evaluation is complete |
Disposition: Level 2, with items 1 through 8 passing. Overall maturity Level 2, with the main gap identified at the Level 4 boundary. Follow-up action: run the correlated-parameter evaluation (item 9) before the next assessment cycle, owner Statistician/CPV analyst, target date six months out. Next scheduled assessment: 12 months from this date.
Common inspection findings this checklist prevents
- A program assumed to be mature because it is old, discovered at inspection to still be setting control limits equal to specification, a Level 1 gap hiding under years of chart history.
- A periodic report and an APR that silently disagree because nobody ever checked them against each other, caught here before an inspector catches it first.
- A correlated-parameter group that has never been evaluated for multivariate monitoring, with no record that the question was ever asked, let alone answered.
- A program with no record of ever having assessed its own maturity, so a self-reported “we have a strong program” has nothing behind it an inspector can check.
How to adapt this checklist
- Set your own assessment cycle in the disposition section; annual is a reasonable default, tied to the same cadence as the CPV plan’s own periodic review.
- Add any item specific to your product’s modality, for example a normalized-metric check for an autologous process, as an additional row rather than replacing an existing one.
- Keep every completed assessment on file across cycles; a single current assessment with no history behind it cannot show whether the program is actually improving.