Opening the book…
An out-of-specification (OOS) result falls outside established specifications or acceptance
criteria. FDA’s OOS guidance addresses chemistry-based testing of drugs within its stated scope, including applicable in-process laboratory tests. Other test types may need different or additional investigation approaches.
A process deviation, an environmental monitoring excursion, or an unusual trend is not automatically the same as an OOS laboratory result. All may need investigation, but the procedure should identify the type of event correctly. An out-of-trend result can remain within specification while differing meaningfully from expected performance.
Examples include a potency result below its approved lower limit, an impurity above its limit, or an identity test that fails its predefined criteria. An extra gel band or a large chromatographic peak is not automatically OOS without knowing the method, expected result, and acceptance criteria. Unexpected observations still require appropriate evaluation.
An OOS investigation should be prompt, thorough, unbiased, scientifically sound, and documented. Do not assume the analyst caused the failure, and do not discard the batch or data as a substitute for finding the cause. The quality unit oversees appropriate investigation and disposition under the quality system.
Initial Laboratory Assessment
Notify the laboratory supervisor promptly and follow the OOS procedure. Preserve relevant samples, preparations, solutions, records, instrument data, and other evidence when practical and safe. Do not continue unplanned testing in the hope of producing a passing result.
The analyst and supervisor review the method, calculations, sample preparation, instrument operation, system suitability, standards, reagents, and relevant records. They also consider sample handling, observations made during testing, training, and possible equipment problems. Document the investigation while evidence is available.
A demonstrated laboratory error may justify invalidating a result, but the explanation must be supported by evidence. A suspected error or a passing retest alone is insufficient. Evaluate whether the identified problem could affect other results and implement appropriate corrective action.
Full Investigation
If the initial assessment does not establish a clear laboratory cause, expand the investigation to manufacturing and other relevant activities. Define the problem, responsibilities,
investigation plan, affected batches, and evidence needed. The quality unit should assess the complete record before deciding disposition.
Review materials, processing and packaging records, equipment, sampling, storage, transport, methods, and relevant trends. Examine whether the event is isolated or could affect other batches or products, including material already distributed. Preserve all relevant results, including failed and invalidated results and the justification for any invalidation.
The investigation report should summarize findings, supported causes or remaining uncertainty, impact assessment, batch disposition, corrective and preventive actions (CAPA), and follow-up. A conclusion of human error alone is inadequate without examining why the error occurred and whether system weaknesses contributed.
Retesting Resampling and Statistical Tests
Retesting uses additional testing of the original sample. Resampling collects a new sample from the batch. Either action needs a scientific rationale and an approved investigation plan. Define the scope and decision rules before obtaining the new results rather than adding tests until the outcome is favorable.
A passing retest does not automatically cancel the initial OOS result. Consider all relevant data and the investigation findings. Resampling may be justified when evidence shows that the original sample was unrepresentative or improperly prepared, but it must not be used simply to replace an inconvenient result.
Do not average an OOS result with passing results to hide variability or create a passing result when the method does not permit that approach. Follow established methods for replicate measurements and report results appropriately.
Statistical outlier tests are not mandatory for every OOS investigation and cannot by themselves establish an analytical error. Their suitability depends on the test and data. They must not be used to discard a valid chemical result solely because it is unusual.
An inconclusive investigation does not prove that the batch meets requirements. The authorized quality unit must make a scientifically justified disposition based on the entire record and applicable obligations. Testing a product repeatedly cannot compensate for an uncontrolled process.
This page titled 14.8: Out of Specification Results was last modified on Sun, 08 Nov 2026 23:35:41 GMT and is shared under a CC BY-SA 4.0 license and was authored, remixed, and/or curated by Emalee MacKenzie.
The book could not be loaded just now.