This is a ready-to-use worksheet. It does not replace a root cause analysis method (fishbone, five whys, fault tree); it is a final writing-quality gate run on the root cause statement after the analysis is done, before it is written into the investigation record. Replace every <<FILL: ...>> placeholder with your own specifics. A filled specimen follows.
Worksheet control header
| Field | Entry |
|---|---|
| Worksheet number | <<FILL: FORM-ID, e.g. FORM-QA-044>> |
| Governing SOP | <<FILL: SOP-ID for deviation / investigation management>> |
| Parent deviation / investigation reference | <<FILL: DEV/INV-ID>> |
| Version | <<FILL: version>> |
1. Draft root cause statement
| Field | Entry |
|---|---|
| Draft root cause statement (as currently written) | <<FILL: paste the exact sentence or passage>> |
| Root cause analysis method used | <<FILL: five whys / fishbone / fault tree / other>> |
| Drafted by (name, date) | <<FILL>> |
2. The three-question defensibility test
Answer each question against the draft statement in section 1. A No on any question means the statement is not yet ready to finalize.
| # | Question | Yes / No | If No, what is missing |
|---|---|---|---|
| 1 | Does the statement explain why the failure was possible as a system property (a missing check, an ambiguous instruction, a design or maintenance gap), not only what the individual did? | <<FILL>> | |
| 2 | Is every claim in the statement traceable to evidence already shown in the investigation’s timeline or evidence section? | <<FILL>> | |
| 3 | Would the corrective action associated with this root cause prevent the same failure from recurring the same way? | <<FILL>> |
3. The “operator error” trap check
Complete this section whenever the draft statement names an individual’s action as the cause, in whole or in part.
| Check | Yes / No |
|---|---|
| The statement names a system condition that made the individual’s action possible or likely (workload, ambiguous instruction, missing independent check, equipment design, fatigue, training gap). | |
| The statement does not stop at naming the action itself (“selected the wrong lot,” “entered the wrong value”) without going on to explain why that action was not caught or was easy to make. | |
| The proposed corrective action is something other than retraining alone (a design change, an added check, a procedure clarification, a system control). |
If any answer above is No, the statement is very likely a symptom description, not a root cause, and should return to analysis.
4. Candidate causes considered
List every candidate cause the investigation considered, whether accepted or ruled out. A root cause statement built from a single considered cause, with no alternatives shown, is a common finding pattern.
| Candidate cause | Ruled in / ruled out | Evidence |
|---|---|---|
<<FILL>> | <<FILL>> | <<FILL>> |
<<FILL>> | <<FILL>> | <<FILL>> |
<<FILL>> | <<FILL>> | <<FILL>> |
5. Confidence and undetermined-cause path
Complete this section only if the root cause could not be conclusively determined.
| Field | Entry |
|---|---|
| Most probable cause, with supporting evidence | <<FILL>> |
| Confidence level (high / moderate / low) and basis | <<FILL>> |
| Monitoring or additional evidence plan that will confirm or rule out this cause | <<FILL>> |
An honest, well-supported “most probable cause, moderate confidence, monitored going forward” statement is stronger and more defensible than a confident but unsupported definitive cause.
6. Disposition
| Field | Entry |
|---|---|
| All three defensibility questions (section 2) answered Yes | Yes / No |
| Operator error trap check (section 3) passed, if applicable | Yes / No / N/A |
| At least one alternative candidate cause documented (section 4) | Yes / No |
| Final root cause statement (revised if needed) | <<FILL>> |
| Reviewed by (name, role, date) | <<FILL>> |
| QA concurrence (name, signature, date) | <<FILL>> |
A No anywhere in this disposition means the statement returns to analysis before it is written into the investigation record.
7. References
21 CFR 211.192 (thorough investigation of discrepancies and failures, including extension to other batches where applicable), the underlying requirement this worksheet’s three-question test is built to satisfy. FDA guidance, Data Integrity and Compliance With Drug CGMP: Questions and Answers (December 2018), on predetermined conclusions and objective evidence.
Confirm the current version of each reference before you rely on it.
Filled specimen
The following shows the worksheet completed for an example deviation, so you can see the level of rigor expected. The company, system, and numbers are illustrative.
Draft root cause statement (section 1): “Analyst entered the wrong dilution factor during sample preparation, causing the reported result to be out of specification.”
| # | Question | Result | Note |
|---|---|---|---|
| 1 | Explains why the failure was possible | No | States only what the analyst did; does not explain why the wrong factor was entered or why nothing caught it. |
| 2 | Every claim traces to evidence | Yes | The wrong entry is confirmed by the LIMS audit trail. |
| 3 | Corrective action would prevent recurrence | No | No corrective action beyond “counsel the analyst” was proposed; would not survive recurrence. |
Operator error trap check (section 3): No system condition named; statement stops at the individual action; proposed action was retraining alone. All three checks failed, confirming this is a symptom description.
Candidate causes considered (section 4):
| Candidate cause | Ruled in / ruled out | Evidence |
|---|---|---|
| Analyst manually keyed the dilution factor with no independent verification step | Ruled in | LIMS workflow confirmed no second-person check exists for this field |
| LIMS calculation template itself was in error | Ruled out | Template formula verified correct against three other results from the same batch |
| Training gap on the dilution procedure | Ruled out | Analyst’s training record current; procedure itself does not require a check |
Revised root cause statement (section 6): “The LIMS workflow for this assay does not require an independent verification of the manually entered dilution factor before the result calculates, so a single keying error was not caught before reporting. Corrective action: add a mandatory second-person verification step for manually entered calculation inputs in this LIMS workflow.”
Disposition: All three defensibility questions now Yes; operator error trap check passed; one ruled-out alternative documented. Reviewed by R. Alvarez, QA, 26 August 2026.
This is the value of the worksheet: the first draft was a fluent, plausible-sounding sentence that named a person and a symptom. It failed the recurrence test because retraining alone would not stop the next keying error. The revised statement, built from the same evidence, points at the missing verification step, a fix that actually closes the gap.
Common inspection findings this worksheet prevents
- A root cause statement that names an individual’s action with no explanation of why the system allowed it.
- A CAPA whose only action is retraining, attached to a root cause that was never actually systemic.
- An investigation narrative that shows a single considered cause, with no evidence that alternatives were examined.
- A confident, definitive root cause claim with no evidence trail supporting it.
- An investigation closed with an unstated or unmonitored “probable cause,” leaving no way to confirm the analysis later.
How to adapt this worksheet
- Set your worksheet number and governing SOP reference in the header.
- Fold this worksheet into your existing investigation record as a mandatory step before the root cause section is finalized, rather than a separate standalone document, if your system prefers a single-document flow.
- Add your own site’s common false-root-cause phrases to section 3 as your reviews identify recurring patterns.
- Route the completed worksheet to the same reviewer who checks the investigation narrative overall, so the defensibility check is not a duplicate review cycle.
- Confirm every reference in section 7 against its current published version before issue.