The practical answer
Keep the material under the appropriate quality status and investigate why the results differ before choosing a preferred number. First determine whether both laboratories tested the same lot, sample population, attribute and reporting basis. A passing retest does not by itself explain an earlier failure or authorize release.
Define the discrepancy and protect the decision
Distinguish a result outside an approved specification from a difference between two results that both comply. The first calls for the established failure procedure; the second may still reveal a method, sampling or trend concern. Record the applicable specification revision and original data rather than revising the limit to fit the result.
For US dietary-supplement operations, 21 CFR 111.113 requires quality-control material review and a documented disposition decision when a specification is not met and in other stated circumstances. Commercial pressure or a supplier's verbal assurance does not replace that decision. [1]
If a specification failure is confirmed, quality control must reject the material unless it approves treatment, adjustment or reprocessing permitted under 21 CFR 111.77. An investigation record alone does not authorize release. [1]
Build a side-by-side evidence record
A supplier certificate and an incoming laboratory report can look contradictory while using different denominators. In an illustrative case, a result of 12.0% on a dry basis corresponds to 11.4% as received when water is 5.0%, provided water is the only basis correction: 12.0 × 0.95 = 11.4. Confirm the actual measurements and permitted calculation before using this explanation.
Do the same check for units, limits and sample identity. A report of less than a quantitation limit is not the same as a measured zero. A retained production sample is not automatically the same sample population as material collected after international transport.
| Comparison field | Question to resolve | Evidence to retain |
|---|---|---|
| Material identity | Are lot, grade and specification revision identical? | Labels, receipt records and supplier lot map |
| Sampling history | Which containers and handling conditions were represented? | Sampling plan, custody record and sample condition |
| Reporting basis | Are units and moisture corrections comparable? | Raw calculation and water or drying method |
| Analytical procedure | Were preparation and measurement equivalent? | Method versions, run controls and relevant raw data |
| Decision boundary | Do uncertainty and rounding affect the interpretation? | Approved decision rule and quality review |
On smaller screens, scroll the table horizontally.
Use additional testing to investigate a hypothesis
State what a proposed experiment could explain before running it. A new preparation of retained sample may examine preparation variability; a documented resampling exercise may examine whether the original sample represented the shipment. They answer different questions. Preserve the relationship between every result and the sample it describes.
FDA's pharmaceutical OOS guidance warns against repeated testing until a passing result appears and distinguishes retesting from resampling. It is a drug-manufacturing guidance, not a substitute for supplement regulations; its scientific reasoning is useful when designing a disciplined investigation. Define the investigation plan and stopping point in advance, and evaluate original and follow-up evidence together. [2]
- Ask the laboratory to review preparation, calibration, dilution, calculations and unexpected observations promptly.
- Preserve remaining sample and relevant data under the applicable retention procedure.
- Agree who authorizes additional work and how it will affect the interpretation.
- Consider whether the issue could affect other lots or finished products using the material.
- Document evidence that supports or rules out each proposed cause; do not substitute speculation for closure.
Close the investigation with an operational change
The useful output is a disposition decision plus any corrective action needed to prevent recurrence. A basis mismatch may require a clearer certificate and specification; a preparation problem may require method training; a genuine lot issue may require supplier action. The conclusion should explain what was learned and why the remaining evidence supports the decision.
When requesting MagneINNO support, provide the lot reference, specification revision, incoming report, method and sample history. A concise evidence package lets the supplier and receiving laboratory investigate the same question without losing time to incomplete comparisons.
Common questions
Can two different results both be correct?
Yes. Different sample populations, moisture bases or analytical definitions can produce different valid results. Establish comparability before treating the difference as a laboratory error.
Should we average the supplier and incoming results?
Do not average unrelated results to make a release decision. First establish their comparability and apply the approved investigation and decision procedure.
Does a new passing sample clear the original failure?
Not by itself. Explain why the original result occurred or how all available evidence affects the lot assessment, then obtain the documented quality disposition required for the material.
Sources & further reading
- 21 CFR 111.113: Material review and disposition decisions
- FDA: Investigating Out-of-Specification Test Results for Pharmaceutical Production, sections III–V