Acceptance Criteria and Out-of-Specification Results

Find the criterion before judging the result

ICH Q6A describes a specification as a set of tests, analytical procedures and acceptance criteria. Criteria can be numerical limits, ranges or other requirements. [A3] A headline such as “99%+” cannot supply all those details. The purity-claim page should be read alongside the relevant lot documents.

For your review, put the specification reference and revision beside the report reference. Check the named attribute, method, units, reporting basis and criterion. If the documents use different terms, resolve the difference before assigning a pass or fail. An HPLC area percentage and a mass-based assay are not interchangeable; identity, purity and assay explains why.

Separate an outside result from an unanswered question

The following fictional examples assume the stated method and reporting basis apply. The numerical comparisons are deliberately simple; they are not recommended acceptance limits.

“Less than 10” includes values on either side of 5. It therefore cannot be rewritten as either “passes” or “fails” that invented limit. A missing result is likewise not a measured failure. Preserve these distinctions in a review log rather than compressing every unresolved item into “OOS”.

For a borderline result, request the laboratory's applicable reporting and decision rules. Do not add rounding, uncertainty allowances or extra decimal places yourself. The figures displayed on a certificate may not contain everything needed to reproduce its conformity decision.

Read the investigation as an explanation

FDA's OOS guidance concerns pharmaceutical production. It calls for documented investigation and does not treat a suspected laboratory error as established without supporting evidence. It also distinguishes the test result from the eventual batch decision. [A4] These are useful questions when reading records; this guide does not assert that those pharmaceutical requirements govern a particular research-material supplier.

A useful review note identifies what happened and what remains unresolved. For example: “Report DEMO-R17 gives 97.8% against the stated 98.0% minimum. The issuer says an investigation is open; no conclusion has been supplied.” This invented wording records the discrepancy without diagnosing its cause or declaring a final release outcome.

Ask for the explanation linking the observed result, the evidence examined and the conclusion. If the issue concerns chromatographic integration, the complete chromatogram and processing context matter; reading an HPLC chromatogram describes the relevant report fields. A replacement PDF without an explanation leaves the reason for the change unanswered.

A retest needs a stated purpose

FDA's guidance distinguishes retesting the original sample material from resampling additional material. It rejects repeated testing until a pass appears, expects a justified plan and requires retention of the original data. Passing retests alone do not invalidate an initial OOS result when no laboratory or calculation error is identified. [A4]

Imagine that a second fictional report gives 98.3% after the original 97.8%. Those two numbers do not tell you whether the same sample was retested, a new sample was collected, a calculation was corrected or an unrelated lot was tested. Do not choose the higher figure or create an average as a substitute for the issuer's documented explanation.

Keep the result and disposition linked

Disposition is the recorded decision about the material. Consult batch release, quarantine and rejection for that distinction. A test result, an investigation in progress and a release decision should not be presented as three names for the same event.

Retain the original report, any revised version and the accompanying explanation together. If you need clarification about a supplied lot, use technical support with the exact identifiers and discrepancy. The quality-documentation hub provides the wider evidence routes. An unavailable procedure is a gap in the material reviewed here, not proof that a supplier has no procedure.