A passing retest is new evidence. It is not, by itself, a reason to delete, discount, or replace the original out-of-specification result.
For finished pharmaceuticals, the regulatory record begins with controls and data. 21 CFR 211.160 requires scientifically sound laboratory controls and documentation at the time of performance. 21 CFR 211.194 requires complete laboratory data, results, calculations, performer identification, and second-person review. 21 CFR 211.192 requires thorough investigation of unexplained discrepancies or failures and a written record of the investigation, conclusions, and follow-up.
FDA's May 2022 OOS guidance for industry gives the Agency's current, nonbinding recommendations for evaluating OOS results. It states that repeated testing until a passing result appears, followed by disregard of the OOS result without scientific justification, is objectionable. It also recommends defining the maximum number of retests in advance, reporting passing and suspect results, and considering all results in the batch decision.
Start with the decision, not the passing number.
The question is not simply whether a later result met specification. The decision question is: What does the complete investigation support about the reliability of the original result and the quality of the batch?
FDA's guidance distinguishes a retest from an automatic override. When clear evidence establishes a laboratory error, the result may be invalidated. When the evidence remains unclear, the firm should conduct a full-scale OOS investigation. The guidance says a discrete result should be invalidated only when an observed and documented test event can reasonably be determined to have caused it.
That means the final record should make the transition visible. If the conclusion changed from “an unexplained OOS exists” to “the original result is invalid” or “the batch remains acceptable,” a reviewer should be able to see the evidence and judgment that support that change.
Define the retest before the answer is known.
A retest plan should test a scientific hypothesis. It should identify the sample, method, number of determinations, acceptance logic, and treatment of the data before new results are produced. If additional testing becomes necessary beyond the established procedure, FDA's guidance recommends a written protocol approved by the Quality unit that describes the testing and the scientific or technical handling of the data.
This order matters because it separates investigation from result selection. A plan created after a favorable result cannot show that the organization chose the test to answer a question rather than to obtain a desired number.
A weak retest record says
- “Retest passed.”
- “Analyst error suspected.”
- “Batch meets specification.”
A reviewable retest record shows
- The hypothesis and evidence that made it plausible
- The approved test design and stopping point
- Every original, passing, and suspect result
- How the combined evidence changed or did not change the conclusion
Keep the original result visible.
FDA's guidance recommends reporting and considering both passing and suspect results in batch release decisions. It also cautions against averaging an original OOS result with additional retest or resample results when that average hides variability. Complete data let the Quality unit evaluate what happened; a selected passing value cannot do that work.
Visibility does not mean every OOS result remains valid forever. It means the record preserves what was observed and separately documents the basis for any later invalidation. This distinction lets a future reviewer tell the difference between an unreliable measurement and an inconvenient result.
Use the Four Lenses to test the record.
The Four Lenses keep the decision components distinct. Each asks a different question about the same OOS investigation.
Evidence
What was actually observed? Preserve the original raw data, calculations, sample identity, instrument and system-suitability information, immediate laboratory assessment, retest protocol, all retest results, manufacturing review, and any related-batch evidence.
Judgment
How were the results interpreted? Show the hypotheses considered, why the retest could test them, why alternatives were accepted or rejected, and why the complete evidence supported invalidation, confirmation, or an unresolved OOS conclusion.
Risk
What uncertainty remained? Address the possibility that the original result reflected batch quality, the impact on other batches or distributed product, and the follow-up needed when the root cause was not fully established.
A practical record example
A finished-product assay result is 96.8 percent against an approved acceptance range of 98.0 to 102.0 percent. The immediate laboratory assessment finds no calculation error, instrument malfunction, or documented departure from the method. The analyst proposes incomplete sample mixing as a hypothesis, but the retained preparation and contemporaneous observations do not establish that it occurred.
Quality approves a predefined retest protocol using retained original sample material. Three new determinations are within specification. The manufacturing review finds no identified process anomaly, and relevant in-process results are within their established limits.
This fictional example does not determine a real batch outcome. It illustrates the record question: what does the organization conclude about the unexplained original OOS, and why? “Three retests passed” is incomplete because the proposed laboratory cause was not established. The Quality record should either document a supported basis for invalidation or explain how the uninvalidated OOS, passing retests, production review, method variability, and remaining uncertainty were considered in the disposition.
| Lens | What the record contains | Question before authorization |
|---|---|---|
| Evidence | Original 96.8 percent result; raw data; method and system checks; three passing retests; production and in-process review. | Are all results and material observations present, attributable, and reviewable? |
| Judgment | Incomplete mixing was considered but not demonstrated; alternative laboratory and manufacturing explanations were assessed. | Does the conclusion exceed what the investigation established? |
| Risk | The original OOS remains unexplained; related-batch and distributed-product impact are evaluated; follow-up is defined. | What uncertainty remains, and what consequence could it have? |
| Authorization | The Quality unit records the batch decision, basis, conditions, responsible approver, and date. | Is the named authority approving the actual conclusion shown by the record? |
Make the conclusion answer four questions.
- What happened? State the original OOS and the scope of the investigation without minimizing or relabeling it.
- What did additional testing establish? Explain what the retest was designed to learn and how each result affected the hypothesis.
- What remains uncertain? Identify unresolved variability, competing explanations, and possible product or related-batch impact.
- What did Quality authorize? Record the disposition and why it was reasonable on the complete evidence available at the time.
A conclusion that answers these questions gives a later reviewer a coherent decision trail. It also makes a weakness visible while there is still time to address it.
The executive conclusion
A passing retest can change an investigation. It cannot erase the event that caused the investigation to begin. Quality leaders should require the record to preserve the original result, define the retest before the answer is known, interpret all results, state the remaining uncertainty, and show the authority behind the final disposition.
To apply the same review to another completed GMP decision, see The Record Should Explain the Decision. For an OOS-specific inspection scenario, review what an FDA investigator may ask about an OOS conclusion.
Source context: Regulatory requirements: 21 CFR 211.160, 21 CFR 211.165, 21 CFR 211.192, and 21 CFR 211.194. Nonbinding agency recommendations: FDA, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, Revision 1, May 2022. The fictional assay and Four Lenses review are CW editorial examples, not regulatory mandates.