How do you investigate an out-of-specification (OOS) result?

1What the rules require.

Under 21 CFR 211.192, “the failure of a batch or any of its components to meet any of its specifications shall be thoroughly investigated, whether or not the batch has already been distributed”. The investigation extends to other batches and products that may be associated with the failure, and its written record includes the conclusions and follow-up.¹ Products that fail their specifications are rejected (211.165(f)), and any deviation from written test procedures must be recorded and justified (211.160(a)).²³ Laboratory records must hold complete data from all tests (211.194).⁴

FDA sets out how to run the investigation in its guidance Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, Revision 1, issued in May 2022. It is guidance, not regulation: it covers chemistry-based laboratory testing of drugs regulated by CDER, it can be used by contract laboratories, and FDA says a contract laboratory’s responsibility is “equivalent to that of a manufacturing firm”.⁵ In the EU, Chapter 6 of the GMP guide expects a written procedure for investigating OOS and out-of-trend results (6.7) and says such data “should be addressed and subject to investigation” (6.9).⁶ MHRA publishes its own expectations as a flowchart guidance, Out of Specification and Out of Trend Investigations (October 2017).⁷

FDA wants the investigation to be “thorough, timely, unbiased, well-documented, and scientifically sound”, and its purpose is to find whether the result comes from the measurement or from the manufacturing process.⁵

2The investigation, phase by phase.

The phases are FDA’s; MHRA splits Phase I into Ia, for obvious errors such as a power failure or spilt sample, and Ib, a checklist investigation by analyst and supervisor, and adds Phase III for the overall review and batch disposition.⁵⁷

3Phase I: the laboratory assessment.

The analyst should check data against the specification before discarding test or standard preparations, keep the preparations if they are stable, and tell the supervisor. Analysts should not continue an analysis they expect to invalidate later for a known error.⁵

FDA lists seven steps for the supervisor’s assessment:⁵

1. Discuss the test method with the analyst and confirm they know and followed the correct procedure.
2. Examine the raw data, including chromatograms and spectra, for anomalous or suspect information.
3. Verify the calculations, and check for unauthorised or unvalidated changes to automated calculations.
4. Confirm the performance of the instruments.
5. Confirm the reference standards, solvents, reagents and solutions were appropriate and met specifications.
6. Evaluate the method’s performance against validation and historical data.
7. Fully document and preserve the record of the assessment.

Hypotheses such as a dilution error or instrument malfunction can be tested on the retained preparations, for example by re-injecting solutions.⁵ MHRA’s checklist adds calibration status, equipment logbooks, system suitability, analyst training, other batches in the same run and, for microbiology, whether plates, controls and incubation were as expected.⁷

“When clear evidence of laboratory error exists, laboratory testing results should be invalidated.” When the evidence is unclear, the investigation moves to Phase II. OOS results “should not be attributed to analytical error without completing an investigation that clearly establishes a laboratory root cause”.⁵ FDA adds that laboratory error should be relatively rare, and frequent errors point to training, equipment or care problems that need corrective action.⁵

4Phase II: production review and additional testing.

The quality unit runs Phase II using a predefined procedure, involving every department that could be implicated and every site involved in manufacture. Its written record should state the reason for the investigation, the parts of the process that may have caused the problem, the documentation review with the actual or probable cause, whether the problem has happened before and the corrective actions taken. It includes the impact on batches already distributed.⁵

Additional testing follows strict rules:⁵

FDA calls repeated testing until a passing result appears, then disregarding the OOS, “testing into compliance”, which is “unscientific and objectionable under CGMP”.⁵ Its laboratory controls Q&A treats a trial injection of a sample to get an unofficial result as a violative practice, because all data from product samples must be retained and reviewed.⁸

5Concluding the investigation.

The quality unit interprets the results and decides the batch. An initial OOS result does not necessarily mean the batch fails. FDA describes three outcomes:⁵

• Invalidated. A test event that can reasonably be shown to have caused the result was observed and documented. The result is not used, but all original data are kept.
• Confirmed. The batch does not meet its specification and is rejected, and the investigation becomes a batch failure investigation extending to other batches and products.
• Inconclusive. No cause is found and the OOS is not confirmed. The result is given full consideration, and release is possible only after a full investigation shows it does not reflect batch quality. The quality unit “should always err on the side of caution”.

“If no laboratory or calculation errors are identified in the first test, there is no scientific basis for invalidating initial OOS results in favor of passing retest results.” All results, passing and suspect, are reported, including on certificates of analysis.⁵ Once a batch is rejected there is no limit to further testing to find the cause, and MHRA adds that a rejection cannot be reversed by it.⁵⁷

For products under an approved NDA or ANDA, 21 CFR 314.81(b)(1)(ii) requires a field alert report within 3 working days of receiving information about any failure of a distributed batch to meet its application specification.⁹ FDA says an initial report should be filed unless the OOS on the distributed batch is found invalid within those 3 days, with a follow-up when the investigation is complete.⁵ In the EU, a confirmed OOS affecting batches on the market from the on-going stability programme should be reported to the competent authorities (6.35).⁶

6How it runs in Seal.

In Seal, an OOS and OOT investigation opens from the result itself, with its sample, raw data, method, instrument and batch already linked. Phase I tasks carry evidence, and a checklist item can be configured so it cannot close without that evidence or a not-applicable rationale. Hypotheses, retest plans, replicates and stopping rules are approved before results are known, and planned testing is reconciled against actual injections, so exploratory work stays visible. The original result is never replaced by a passing retest, and batch disposition is a separate, attributed decision: closing a laboratory task cannot release material.

Neil can gather the evidence for an OOS investigation, bringing the original result, preparation records and instrument evidence into a brief with targeted evidence requests. The laboratory supervisor assesses whether an assignable cause is documented and QA approves any invalidation. In the LIMS, the investigation opens on the same records as the result it concerns.