An out-of-specification result is a fact requiring investigation. It is not automatically a laboratory error, and it does not disappear because a repeat test passes.
The investigation must preserve the original result, assess possible laboratory causes promptly and expand beyond the laboratory whenever no clear, scientifically supported assignable cause is demonstrated. The purpose is to understand the result and its impact—not to test until a desirable answer appears.
1. Preserve evidence and take immediate action
Notify the responsible laboratory supervisor and Quality function promptly. Preserve preparations, solutions, standards, chromatograms, instrument files, audit trails, worksheets and relevant equipment state where feasible. Record what was observed before memory and evidence change.
- Confirm the reported result and applicable specification.
- Check calculations, units, transcription and sample identity.
- Protect remaining sample and preparations according to stability limits.
- Identify potentially affected batches, materials or ongoing operations.
- Do not discard, overwrite or invalidate data before authorised assessment.
2. Conduct a focused laboratory assessment
The analyst and supervisor should reconstruct the execution against the approved method. Review raw data, system suitability, sample and standard preparation, reagents, instrument performance, environmental conditions and deviations from the method.
| Area | Questions |
|---|---|
| Method execution | Were every step, time, temperature, dilution and calculation followed? |
| Materials | Were standards, reagents, columns and consumables correct and within validity? |
| Equipment | Were calibration, maintenance, alarms, sequence events and audit trails acceptable? |
| Sample | Was sampling, storage, preparation, homogeneity and stability appropriate? |
| Data | Are all injections, integrations, repeats, aborted runs and metadata retained? |
| Analyst interview | What was observed at the time—not what is assumed after seeing the result? |
3. Test hypotheses—do not perform exploratory retesting
Additional laboratory work should answer a documented hypothesis. For example, suspected incomplete extraction may be challenged using retained preparation and a predefined comparison. A passing result alone does not prove the hypothesis.
State the suspected cause, predicted observation, experimental plan and acceptance logic before performing the test. Preserve every result, including those that contradict the hypothesis.
4. Control retesting and resampling
Retesting uses another portion of the original homogeneous sample; resampling obtains a new sample from the batch. Each answers a different question and requires an approved, scientifically justified plan. Predetermine the number of tests and how all results will be evaluated. Averaging an OOS result with passing retests can hide variability and is generally inappropriate when individual results must meet a specification.
- Never use repeated testing to obtain a passing result.
- Use qualified analysts and approved methods.
- Consider sample homogeneity and sampling error separately.
- Do not replace the original result without a demonstrated invalidating cause.
5. Expand to manufacturing and batch history
If the laboratory assessment does not demonstrate a clear assignable cause, investigate production, sampling and product history. Review process parameters, materials, equipment, hold times, deviations, environmental data, other tests, stability, complaints and related batches.
- Assess whether the result reflects real process or product variability.
- Review trends rather than treating the result in isolation.
- Evaluate other batches made with common materials, equipment or conditions.
- Consider whether the specification, method or process capability needs attention.
6. Reach and document the conclusion
| Conclusion | Evidence expectation | Typical consequence |
|---|---|---|
| Confirmed laboratory error | Specific cause demonstrated and original result scientifically invalidated. | Correct laboratory cause; evaluate valid replacement data. |
| Manufacturing or product cause | Evidence connects result to process, material, sampling or product failure. | Assess batch and related scope; investigate and implement CAPA. |
| No assignable cause | Investigation cannot scientifically invalidate the result. | Original OOS remains relevant to batch evaluation. |
| Inconclusive investigation | Evidence or sample limitations prevent firm cause determination. | Document uncertainty, product risk and further action; do not convert uncertainty into a pass. |
Common OOS investigation mistakes
- Starting retests before documenting the initial assessment
- Assuming analyst error without evidence
- Testing into compliance
- Discarding failed or aborted injections
- Using a passing retest to invalidate the original result
- Failing to expand outside the laboratory
- Treating resampling as interchangeable with retesting
- Writing the conclusion first and selecting supporting data afterward
Primary references
Accessed 23 September 2026. Verify current requirements applicable to your products, jurisdiction and approved pharmaceutical quality system.
Educational notice: This article provides general education and does not replace approved procedures, product-specific scientific assessment or regulatory requirements.