Out-of-Specification Incoming Raw Materials: The Two-Phase Investigation Process GMP Auditors Want to See

An out-of-specification result on an incoming raw material is inconvenient. An out-of-specification result with no documented investigation — or one that was clearly reverse-engineered to justify retesting — is a regulatory crisis. And given how consistently this pattern appears in FDA 483 observations and Warning Letters, a lot of manufacturers still haven’t sorted out the difference.

The test result is the signal. What happens next is the test of your quality system.

What GMP Actually Requires When a Raw Material Fails

For pharmaceutical manufacturers, the baseline is FDA’s 2006 Guidance for Industry: Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production. For dietary supplement manufacturers operating under 21 CFR Part 111, the OOS guidance is less prescriptive — but §111.75, §111.77, and §111.95 together establish an unambiguous expectation: when a raw material doesn’t meet its specification, you investigate before you decide what to do with it. Not after. Not simultaneously. Before.

ICH Q7 (Good Manufacturing Practice Guide for Active Pharmaceutical Ingredients) reinforces this for API suppliers, requiring that “the cause of the OOS result shall be determined” and that investigations be documented, completed within a defined timeframe, and formally reviewed by the quality unit. Record retention under §111.95 extends to 1 year beyond the product’s expiry date — which means an OOS investigation file from a batch manufactured today may need to survive a GMP audit years from now.

USP chapter <1010> (Analytical Data Interpretation and Treatment) draws the hardest line of all. It distinguishes sharply between valid retesting — which can only follow a Phase I investigation that has identified a confirmed, assignable laboratory cause — and invalid retesting, which is simply running the analysis again hoping for a different number. The latter is common in practice. It is not permitted under GMP, regardless of whether the retest passes.

Phase I: Start in the Laboratory, Not at the Receiving Dock

The two-phase investigation framework starts in the analytical testing laboratory, not in the warehouse. This sequencing matters because, in practice, a significant portion of OOS results for incoming raw materials are laboratory events rather than genuine material failures. Skipping Phase I and moving straight to material disposition means you may be rejecting — or worse, accepting — material based on a flawed analytical result.

Phase I has a defined scope: determine whether the OOS result can be attributed to an identifiable, documented error in the analytical process. This covers four main areas.

Sample preparation. Was the sample weight within the method’s specified range? Was the correct extraction solvent used at the right concentration and temperature? For botanical raw materials in particular, incomplete extraction is a frequent source of artificially low potency results. A 5–10% deviation in extraction solvent volume can produce apparent potency results that fall just below specification for concentrated extracts.

Instrument performance. Did the instrument pass its daily system suitability checks before and during the analytical run? For HPLC-based methods, the tailing factor, theoretical plate count, and percent relative standard deviation (%RSD) for replicate standard injections all need to be within specification. An instrument that passes system suitability can still produce erroneous peak integrations if the software’s baseline correction settings haven’t been validated for the matrix being tested.

Reference standard integrity. Degraded or incorrectly stored reference standards will produce false OOS results across every sample in the run. A thorough Phase I investigation checks the standard’s certificate of analysis, storage temperature logs for the preceding 30 days, and expiry date. If the standard has been subjected to freeze-thaw cycling or stored outside the specified humidity range, the Phase I conclusion should reflect that.

Analyst documentation. If the analyst’s notebook doesn’t record the time of sample preparation, the reconstitution conditions, and the injection sequence with timestamps, you cannot rule out technique-related variance. This is where documentation discipline creates or destroys your ability to conduct a defensible investigation.

Phase I should be completed within 30 business days of the original OOS result. The FDA’s 2006 guidance establishes this benchmark for pharmaceutical manufacturers, and GMP auditors apply it broadly to supplement facilities as well. Document every step — including the steps that find nothing wrong. “Phase I investigation performed; no assignable cause identified” is a fully legitimate conclusion. “Investigation not completed” is not.

If Phase I confirms a laboratory error, retesting is authorized. The analyst who performed the original test should not perform the retest. That’s a structural control, not a soft recommendation, and it’s explicitly addressed in FDA’s 2006 guidance.

Phase II: When the Investigation Moves to the Material

If Phase I finds no assignable laboratory cause, you move to Phase II: an investigation into whether the raw material itself genuinely fails its specification.

This is where sourcing history becomes directly relevant. A longstanding supplier with 20 consecutive conforming lots is a different risk profile from a new overseas manufacturer providing their third shipment. Phase II documentation should reflect that context — including what you knew about the supplier at the time of the OOS.

A complete Phase II investigation addresses five areas:

COA cross-examination. Does the supplier’s reported value for the failed parameter actually align with your incoming specification? COA transcription errors — where a supplier’s own documented result is technically outside your spec but was never caught at receiving — are more common than most QA teams want to acknowledge. If you catch one at this stage, that’s a specification management problem, not a supplier quality problem.

Chain of custody and sampling review. Was the correct lot sampled? Was the sampling tool clean and dry? For bulk powder raw materials, cross-contamination between lots during receiving is a real failure mode, especially when samples are pulled in a shared receiving area with tools that aren’t dedicated per-lot. A sampling error doesn’t mean the material fails; it means the result doesn’t tell you anything.

Historical trend analysis. Pull results for the previous 3–5 lots of this material from the same supplier and plot the failed parameter over time. A gradual downward trend in potency across multiple lots suggests the supplier may have changed their manufacturing process or raw material source. A single-lot failure against a flat historical trend is a different kind of problem, often more containable and more likely to be a sampling or analytical artifact.

Extended testing. Depending on the failed parameter and the disposition stakes, Phase II may include sending a retained sample to an independent analytical testing laboratory for confirmatory analysis. This step adds 3–7 business days to the investigation timeline, but it substantially strengthens a rejection decision — especially when a supplier disputes the results. A third-party result from an ISO 17025-accredited analytical testing laboratory carries weight that in-house data alone often doesn’t in a supplier dispute.

Disposition decision and documentation. At the end of Phase II, the quality unit makes a formal disposition decision: accept under a documented exception, reject, or place on hold pending additional data. This decision requires approval signatures and must be entered in the batch or lot record. Informal disposition — material quietly moved from quarantine to production without a signed quality decision — is both a GMP violation and a traceability failure.

Building a CAPA That Survives the Next Audit

An OOS investigation without a corrective and preventive action plan is an incomplete quality event. The CAPA needs to address the root cause — not the OOS result itself, but the system failure that allowed a non-conforming material to reach the testing stage without earlier detection.

Root cause statements are where many QA teams underinvest. “Analyst error” is not a root cause. “Analyst failed to equilibrate the column for the required 30 minutes prior to sample injection, as evidenced by retention time drift of approximately 0.3 minutes across the first 8 injections of the run” is a root cause. The difference between these two statements determines whether your CAPA will actually prevent a recurrence — or whether the same OOS will appear on your next audit’s observation list.

Corrective actions worth considering, depending on root cause:

• For supplier drift: tighten incoming specification limits or add a pre-shipment testing requirement, directing the supplier to use an accredited third-party lab before materials are released for shipping
• For sampling technique variability: update the sampling SOP with explicit tool-cleaning steps, per-lot sampling equipment, and a second-person verification for high-risk materials
• For reference standard management: implement a dedicated standard log with storage condition monitoring, expiry tracking, and a mandatory standard verification step before each analytical run
• For documentation gaps: revise the analyst notebook template to require time-stamped entries at each preparation step

The preventive action — the “PA” in CAPA — is where the quality system either learns or doesn’t. A corrective action that addresses only this lot improves nothing systematically. Auditors know this, and during a follow-up inspection they will specifically ask what changed to prevent the same OOS from occurring on the next incoming shipment of this material.

Five Mistakes That Convert an OOS Into an Audit Finding

These show up repeatedly in 483 observations and Warning Letters across both pharmaceutical and supplement GMP inspections:

1. Retesting before Phase I is complete. Even if the retest result passes, unauthorized retesting creates a documentation problem that’s harder to explain than the original failure.

2. Averaging the OOS result with subsequent passing results. Under USP <1010> and FDA’s 2006 guidance, averaging a confirmed OOS with retest values to calculate an “acceptable” mean is explicitly not permitted.

3. Vague root cause language. “Analyst error” or “equipment malfunction” without specific, evidence-based detail will not satisfy an auditor — and won’t prevent recurrence.

4. Open CAPAs at the time of the audit. A CAPA with a due date that has passed without closure is a finding on its own, independent of the underlying OOS. If the due date needs to change, document the justification and get quality unit approval before the deadline passes.

5. Missing the 30-day investigation window. Investigations that extend past 30 business days without documented justification suggest either that the quality system lacks the capacity to manage OOS events or that the investigation is being deliberately delayed. Neither impression is helpful during an audit.

The investigation is where most manufacturers fail. The OOS result itself is often a manageable quality event — even a complete lot rejection, handled properly, demonstrates a functioning quality system. The absence of a documented, time-bound, root-cause-driven investigation is what turns a routine quality event into a regulatory observation, and in repeat-inspection situations, potentially something more serious.

---

Written by Nour Abochama, VP Operations, Qalitex | Quality Consultant, Ayah Labs. Learn more about our team

Talk to our team about raw material testing. Contact us

Related from our network

• Third-Party Raw Material Verification and ISO 17025-Accredited Testing at Qalitex Laboratories — Independent confirmatory testing for OOS investigations, incoming material verification, and supplier qualification support.