Worksheet: OOS Retest and Resample Statistical Decision

This is a ready-to-use worksheet. It does not replace the OOS investigation record; it is the working paper that documents the statistical and scientific reasoning behind one specific retest, resample, or invalidation decision, and it is attached to the investigation record as supporting evidence. Replace every <<FILL: ...>> placeholder with your own specifics. Complete Part C and obtain QA authorization before any new result is generated; completing it afterward, once the numbers are known, is exactly the practice this worksheet exists to prevent. A worked filled specimen follows. Verify each cited regulation and statistical method against the current source before you rely on it, and treat this as an educational reference to adapt rather than as statistical, legal, or regulatory advice; consult a qualified statistician for your actual data.

Document control header

Instructions: when and how to use this worksheet

Use this worksheet at the point Phase 1 has closed with no assignable laboratory cause, or wherever the investigation needs a documented, scientifically reasoned choice between retesting, resampling, and accepting a result as confirmed. Three principles govern every entry:

1. The plan comes before the data. Part C is completed and signed by QA before a single additional preparation, retest, or resample is analyzed. An evaluation rule chosen after seeing the numbers is not a plan, it is the definition of testing into compliance.
2. The original result never leaves the evaluation. Retesting or resampling adds information; it does not delete the original reportable result from the record or from the final statistical judgment. A defensible worksheet shows the original value sitting in Part D next to every new value, not filed away once a passing number appears.
3. Averaging more values together is not a way to make a number smaller. Growing the number of preparations that feed a mean narrows the spread around that mean and pulls extreme values toward the middle, whether or not the extreme value is the true one. That is exactly why the plan, the number of additional results, and the rule for judging the combined set are fixed in advance. Consult USP General Chapter <1010>, Analytical Data, Interpretation and Treatment, for the current statistical methods applicable to replicate measurements and outlier evaluation; this worksheet structures the documentation, it does not substitute for that chapter or for statistical expertise on the specific data.

Part A: Result and eligibility check

Part B: Retest, resample, or accept, decision logic

Select exactly one path and record why the other two do not apply. Do not select a path because it is the one most likely to produce a passing result.

Part C: Predefined statistical evaluation plan (complete and sign before any new result exists)

Part D: Execution results and statistical evaluation

Complete only after Part C is signed and testing is performed. List every individual value, not only a combined mean.

Part E: Decision and sign-off

Acceptance criteria

This worksheet is complete and defensible when all of the following are true:

• Part C is signed and dated before Part D contains any result.
• Every individual value is listed in Part D, not only a combined mean.
• The original reportable result appears in the evaluation in Part D and is never dropped from it.
• Any outlier test applied was pre-approved in Part C, is never applied to remove the result of interest from a chemical assay, and is tied to a validated procedure where used for a biological assay.
• The final decision in Part E follows directly from the predefined rule and the data, not from which outcome is more convenient.

Retention

Retain this worksheet with the linked OOS investigation record for not less than <<FILL: retention period>>, per the records retention schedule. It is part of the batch record chain where the linked OOS affects a distributed or in-process batch.

References

21 CFR 211.192 (investigation of unexplained discrepancies and failures to meet specifications). 21 CFR 211.194 (laboratory records and the underlying raw data). FDA Guidance for Industry, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production (originally October 2006, Level 2 revision May 2022), for the retest, resample, and statistical-evaluation principles this worksheet documents. USP General Chapter <1010>, Analytical Data, Interpretation and Treatment, for the applicable statistical methods for replicate measurements and outlier evaluation. Describe and apply per the current chapter; do not copy its text into site procedures.

Confirm the current version of each reference before issue.

Filled specimen

The following shows the worksheet completed for a resample decision on a raw material assay, so you can see the level of detail an inspector expects. The company, material, and numbers are illustrative; replace them with your own.

Part A:

Part B:

Part C:

Part D:

Part E:

Note what makes this defensible: the reason for resampling was on file (DEV-2026-0210) before the new sample was pulled, the evaluation rule did not depend on the resample passing, and the original value was explained, not erased. A resample pulled first and justified afterward is the same defect as a retest evaluated after the fact.

Common inspection findings this worksheet prevents

• Retest or resample decided, or the acceptance rule chosen, after the new results were already known.
• A combined mean reported without the individual values behind it, hiding that most of the individual results still fail.
• An outlier test applied to a chemical assay to remove the one result that failed.
• A resample performed with no documented reason the original sample was unrepresentative, or that reason written up after the resample already passed.
• The original OOS quietly dropped from the record once a later result passes.

How to adapt this worksheet

1. Set the worksheet number and the governing SOP in the header to match your OOS procedure.
2. Where your reportable result is a mean of multiple preparations, extend Part D’s value list to however many individual determinations the method defines, and state the reportable-result formula in Part A exactly as the method does.
3. If your site uses a validated outlier procedure for a specific biological assay, name the procedure and the assay types it applies to in Part C, and keep the “never for the result of interest in a chemical assay” rule intact.
4. Confirm every regulation and the USP chapter reference in the References section against the current published version before issue.