When Your Raw Material Fails Testing: What 21 CFR Part 111 Actually Allows — and What It Doesn't

A kilogram of elderberry extract arrives at your Chicago warehouse. Your analytical testing lab reports a potency failure — anthocyanin content comes in 14% below your specification. The supplier insists it’s a lab error. Your QA manager wants to retest. And someone in procurement quietly suggests you can “just use it this time, we’re behind schedule.”

Before any of those decisions get made, there’s a regulatory framework that governs exactly what happens next. If your written procedures don’t reflect that framework, you’re building quiet risk into every batch record you’ve ever created — and every one you’ll create going forward.

What 21 CFR Part 111 Actually Requires at the Point of a Raw Material Failure

The dietary supplement GMP rule — codified at 21 CFR Part 111 since the FDA’s 2007 final rule published at 72 FR 34752 on June 25, 2007 — doesn’t leave much room for ambiguity here. Under 21 CFR 111.80, if a component fails to meet a specification, you must either reject it outright or follow a pre-existing written procedure that addresses the failure. Not a verbal agreement with the supplier. Not a note in the margin of the packing slip. A pre-existing, approved, written procedure.

That same framework requires you to have specifications for every component before you receive it (21 CFR 111.70(b)) and to test against those specifications before the component is used in production (21 CFR 111.75(a)(1)). The spec → test → decision chain isn’t optional infrastructure — it’s the backbone of a defensible incoming QC program.

What catches brands off guard is the follow-up question: what does “follow a written procedure” actually permit? This is precisely where “retesting” becomes either a legitimate quality tool or a compliance problem, depending entirely on the structure and documentation around it.

Three Circumstances That Justify Additional Testing — and Two That Don’t

There’s a meaningful distinction between retesting (running the original sample through the same method a second time) and resampling (pulling a fresh specimen from the lot under a documented sampling plan). Both can be appropriate. Neither is appropriate without a documented rationale that precedes the additional test.

Legitimate circumstance 1: An assignable cause is identified during phase 1 investigation

If a structured investigation — carried out by your QA team or, in the case of third-party work, documented in writing by your contract analytical testing laboratory — reveals a specific, identifiable cause for the failure, retesting with that error corrected is scientifically and regulatorily sound. Assignable causes include documented instrument malfunction, verified analyst preparation error, a confirmed reagent lot failure, or a sample handling deviation with supporting records. The cause must be identified and documented before additional testing begins, not inferred after the retest happens to pass.

Legitimate circumstance 2: Sampling error in an otherwise homogeneous lot

Botanical powders and extracts often have real variation within a lot, particularly near container walls, headspace, or unmixed settling layers. If your sampling protocol failed to account for that — say, a single draw from the top of an agitated container — a properly documented resampling from multiple representative points within the same lot can be appropriate. Your written procedure should define the sampling criteria and how many additional samples are permitted.

Legitimate circumstance 3: Confirmatory testing under a pre-established protocol

If your specification was validated against a method with documented inter-analyst variability, and your SOP includes a provision for confirmatory testing by a second analyst or a second qualified laboratory, running that confirmatory analysis is defensible. But “confirmatory” means exactly what it says — it tests whether the original result is reproducible, not whether you can find a number you prefer.

Where it stops being legitimate:

If your phase 1 investigation finds no assignable cause, you cannot simply retest until you get a passing result. This pattern — informally known in the industry as testing into compliance — is one of the more telling GMP red flags in dietary supplement inspections. FDA has issued warning letters to supplement manufacturers specifically for conducting multiple rounds of retesting on failing raw materials without a documented phase 1 finding to support each round. The batch records in those cases showed repeated attempts, no investigation reports attached, and eventual release on a passing result.

FDA’s position has been consistent: a failing result is a data point, not a coin flip. Repeating the test without addressing the question why it failed doesn’t erase the original result. It adds to the record.

The same principle applies to numerical averaging. Averaging a failing result with a passing one doesn’t produce a compliant result — it produces a laundered number that obscures a real signal in your raw material.

What Your OOS Written Procedure Must Actually Address

This is the gap that appears most frequently in 483 observations at supplement facilities: a company has an OOS procedure on paper, but it was drafted for finished-product testing and never properly adapted for incoming raw material failures, or it’s silent on the conditions that limit when additional testing may be performed.

A raw material OOS SOP that will hold up to scrutiny needs to address, at minimum:

Scope and decision triggers. The procedure should specify exactly which test systems and which types of failures it governs. An anthocyanin potency failure is handled differently than a heavy metals exceedance, which is handled differently than a microbiological limit failure. One generic procedure often isn’t enough.

Phase 1 investigation scope and timeline. Define who conducts it, what records are reviewed (instrument maintenance logs, analyst qualification records, sample handling documentation, reagent lot history), and what timeline is required for completion. In practice, 3–5 business days is a reasonable standard for an incoming raw material investigation. Longer stretches with no documented progress are themselves a records issue.

Explicit limits on additional testing. The procedure should state clearly — in writing — how many additional tests may be run under a phase 1 finding, what documentation is required for each, and what the final disposition criteria are. Procedures that read “additional testing may be conducted at management discretion” are the kind of open-ended language that generates 483 observations, because “management discretion” isn’t a documented control — it’s an undocumented decision.

Disposition options at each decision point. Reject, quarantine pending full investigation, or conditional release with documented justification. Each path should be defined with the authority level required to execute it.

Documentation requirements — including negative findings. Every step of the investigation must be recorded, whether it finds something or doesn’t. A phase 1 investigation that concludes “no assignable cause found” is a valid conclusion. It just means you’ve confirmed you have a genuine raw material failure, not a lab error, and your disposition decision should reflect that.

That last documentation point matters more than most quality teams initially appreciate. An ISO 17025-accredited analytical testing lab will report every result, including failures, with full traceability to instrument calibration records and analyst qualifications. There’s no mechanism to request an informal preliminary result before the official report. That design is intentional — it protects the integrity of the measurement. Your internal raw material testing program should operate under the same principle.

Building This Into Your Quality System Before You Need It

The worst time to write an OOS procedure is after you have a failing result in hand. Under time pressure — and there’s always time pressure when a production run is scheduled — the procedure tends to be shaped to fit the immediate situation rather than to govern it. That’s precisely the kind of document FDA investigators are trained to read critically.

A few practical steps worth implementing before you need them:

Set specifications before you order the material. Under 21 CFR 111.70(b), specifications must exist before a component is received. If you don’t have a written specification for anthocyanin content in your elderberry extract, you can’t have a meaningful pass/fail result — and you can’t have a legitimate OOS investigation either. That sounds obvious, but new ingredient introductions often skip the spec-setting step until the first shipment is already in quarantine.

Define your acceptable retesting protocol in writing, with explicit limits. How many additional tests may be run in phase 1 (typically none without an assignable cause)? How many under a documented phase 2 investigation? What is the final disposition criterion if phase 2 also fails? An SOP that doesn’t answer these questions isn’t a controlled procedure — it’s a permission slip.

Ensure your contract lab is accredited and communicates directly with QA. If you’re sending raw material samples to a third-party analytical testing lab, that lab’s ISO 17025 accreditation means their results are traceable, reproducible, and fully documented. But accreditation alone doesn’t guarantee your results get to the right person at your company. Specify in your testing agreement that OOS results are communicated directly to your QA team — not only to the purchaser or the logistics contact who placed the sample submission.

Run a simulated OOS drill at least once a year. Introduce a hypothetical failing result into your incoming QC process and walk your team through the full investigation procedure in a low-stakes environment. It sounds like administrative overhead until the first time you have a real failure under time pressure, the production team is waiting, and your QA team is reading the SOP for the first time. Knowing where the procedure is ambiguous before it matters is worth considerably more than discovering it mid-event.

FDA issues roughly 30–50 dietary supplement warning letters per year, and 21 CFR Part 111 GMP citations are among the most consistently observed findings. The cases that escalate into consent decrees or injunctions almost always involve a pattern of failures over time, not a single isolated event. An OOS procedure that lets failing results slip through without documented investigation is the kind of recurring pattern that creates that record.

Getting ahead of it with a solid written procedure is one of the lowest-cost, highest-return investments in your quality system — and one of the few that your batch records will actually show.

---

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

Ship your sample to our Chicago facility — get a Qalitex CoA in 5–7 days. Contact us

Related from our network

• ISO 17025-Accredited Supplement Testing at Qalitex Laboratories — Third-party analytical testing for raw materials and finished dietary supplements, with full CoA documentation under ISO 17025 accreditation.