OOSOut Of Specification

An Out-of-Specification (OOS) result is any laboratory test result that falls outside the specifications or acceptance criteria established in the new drug application (NDA / ANDA), the drug master file, the official compendium (USP, EP, JP) or the drug manufacturer's release / stability specifications. OOS applies to release tests, stability tests and in-process tests that have approved specifications. It can occur on an active pharmaceutical ingredient, a finished drug product, an excipient, a stability time-point or a process-validation sample.

OOS is a specific term inside a broader family. Out-of-Trend (OOT) results are within specification but outside the expected statistical range (e.g. a stability assay that drops faster than the historical trend). Out-of-Expectation results are within specification but anomalous in some other way. Atypical results are within specification but the analyst has reason to doubt them. All three categories require investigation; only OOS triggers the formal Phase 1 / Phase 2 workflow the FDA's 2006 guidance defines.

The FDA's October 2006 'Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production' guidance is the authoritative US framework. It is short (about 12 pages), prescriptive and surprisingly readable. Its core structure is a two-phase investigation:

1. Phase 1 — Laboratory investigation. Conducted by the QC laboratory before any results are reported beyond the lab. Determines whether the OOS was caused by a laboratory error (analyst, instrument, method, reagent, sample-handling).
2. Phase 2 — Full-scale investigation. Triggered when Phase 1 does not identify and confirm a laboratory error. Conducted by the Quality Unit, extends into manufacturing, and follows 21 CFR 211.192.

The guidance was reissued in essentially the same form in 2010 (replacing references to the withdrawn draft 1998 guidance with the 2006 final), and remains active. Subsequent FDA pronouncements (data-integrity Q&A 2018, warning letters citing the guidance) have if anything stiffened the expectations.

The analyst's first action on observing an OOS is to inform the supervisor and preserve the original test solution, the original injection, the original calculation and the original chromatographic data (if applicable). The next action is not to retest. The 2006 guidance is explicit: 'retesting may be conducted only after the initial assessment by the supervisor, and only as part of an OOS investigation, never as a replacement for the original OOS result.'

The Phase 1 checklist

1. Discuss the testing method, calculations, instrument operation, system suitability and analyst observations with the analyst.
2. Examine the data — chromatograms, calculations, system suitability, calibration of the instrument used.
3. Examine the sample preparation — weighing, dilution, transfer, filter, sonication, vortex.
4. Examine the reagents and reference standards — lot, expiry, qualification.
5. Examine the instrument — calibration status, system suitability, baseline noise, retention-time drift.
6. Determine whether an obvious laboratory error explains the OOS (wrong weighing, wrong dilution, wrong injection volume, leaking column connection, expired standard).
7. If a laboratory error is identified and confirmed, invalidate the original result with full documentation and re-test. The invalidated result, the reason, and the re-test result all remain in the lab record.
8. If no laboratory error is identified, escalate to a Phase 2 investigation.

If Phase 1 does not identify a definitive laboratory error, the Quality Unit takes over and initiates a full-scale investigation under 21 CFR 211.192. This investigation has the following obligations:

• Review the manufacturing history of the batch (BMR, deviations, equipment performance, environmental data, in-process results, raw-material lots).
• Review the batch's sample-handling history (sampling SOP, sample container, transport, storage, chain of custody).
• Extend the investigation to other batches of the same drug product, and other drug products that may have been associated with the failure (the famous 'shall extend' clause).
• Decide on retest, resampling and other additional testing in accordance with a pre-approved investigation plan, not ad-hoc.
• Conclude with a written investigation report signed by the Quality Unit.

Retest vs resample

The 2006 guidance carefully distinguishes retest from resample. A retest uses the original sample (or a portion of the original sample) and is appropriate when the lab investigation suggests a result-specific issue. A resample takes a new sample from the original batch material and is appropriate when the original sample may not be representative. Both are decided in advance, executed against a documented plan, and the results — including any further failures — remain in the record. The guidance is clear that 'when no laboratory or statistical outlier rationale can be identified, the OOS investigation should not be terminated, and an outlier test should not be used to invalidate the result.'

USP <1010> covers statistical treatment of analytical data and includes guidance on outlier tests (Grubbs, Dixon and others). The FDA's position in the 2006 guidance is that outlier tests have a narrow legitimate role: 'an outlier test may be of some value as part of the evaluation of the significance of that result.' But they cannot be used to invalidate an OOS result without an identified, scientifically valid assignable cause.

In practice this means biological assays (with high inherent variability and pre-validated outlier procedures) may use outlier tests as part of release calculation. Chemical assays (with low inherent variability) cannot use outlier tests to discard inconvenient values. The decision must be made in the method-validation phase, not at the moment an OOS appears.

Two recurring failure modes in OOS handling:

Averaging away the failure

Reporting the average of two injections when one is OOS and the other is in-spec, without first investigating the disagreement, is a violation of the 2006 guidance. Averaging is valid only when the method's validation explicitly supports it (typical for HPLC assay: two injections, the result is their mean) and the individual injections are within method-defined repeatability. If the individual injections disagree by more than the validated tolerance, the disagreement itself is an OOS investigation trigger.

Rounding before comparison

If the specification is NLT 98.0% assay and the reported result is 97.95%, the rounding question matters. The convention recorded in the method validation (and consistent across all batches of the product) governs — typically result is reported to the same precision as the specification, with normal rounding (97.95 → 98.0, OOS not triggered). Inconsistent rounding between batches, or invoking rounding only when it rescues an OOS, is a finding.

OOS on a stability time-point has additional considerations because the failure may indicate a shelf-life issue, not just a single-batch problem. The investigation must address: was this batch unusual; do other batches at the same time-point show the same trend; is the labelled shelf life still defensible; does the manufacturer need to file a stability-failure report (Field Alert Report under 314.81(b)(1)(ii) for NDAs).

Field Alert Reports under 21 CFR 314.81 are required within three working days for any 'information concerning any incident that causes the drug product or its labeling to be mistaken for, or applied to, another article' and within three working days for 'information concerning any bacteriological contamination, or any significant chemical, physical, or other change or deterioration in the distributed drug product, or any failure of one or more distributed batches of the drug product to meet the specification established for it in the application.' A stability OOS on a distributed batch may trigger a FAR.

1. Original OOS invalidated without documented assignable cause — the Barr Decision pattern.
2. Retesting until a passing result is obtained without a pre-approved investigation plan.
3. Phase 1 investigation done by the analyst without supervisor involvement; lab error 'identified' to make the OOS go away.
4. Phase 2 investigation did not extend to other batches as 211.192 requires.
5. Outlier tests used on chemical assay data to discard the OOS value without an assignable cause.
6. Averaging invoked to mask a discordant individual injection that was itself an OOS.
7. Chromatographic data deleted, reintegrated, or reprocessed without audit trail — data-integrity finding stacked on top of OOS finding.
8. Investigation closed without a Quality Unit signature on the disposition.
9. Stability OOS not triggering a Field Alert Report when 314.81 applied.
10. Pattern of OOS at the same instrument, same method, same analyst not trended.

The MHRA's 2018 GxP data-integrity guidance singles out OOS handling as a high-risk area for integrity violations because of the operational pressure to make the result go away. The data-integrity controls that matter most around OOS:

• Every chromatographic injection — including 'trial' injections — is retained in the original system with audit trail. The 'inject and look before saving' anti-pattern is forbidden.
• Re-integration of a chromatogram is captured in the audit trail with reason for change. The original integration is preserved.
• Reprocessing of raw data uses validated parameters; ad-hoc parameter changes to rescue a result are flagged in the audit trail.
• The analyst cannot delete an injection, a sample, a sequence or a result. The system enforces append-only.
• Two-person review of OOS investigations is the rule, not the exception.

• Pre-approved Phase 1 / Phase 2 investigation SOP with checklists and pre-approved retest plans for common scenarios.
• Mandatory supervisor involvement at the moment of OOS observation — supervisor and analyst initial together before any further action.
• OOS is its own record type, linked to the originating sample, batch, instrument, method, analyst and chromatographic data file.
• Quality Unit independence from the lab on Phase 2 — the lab investigates Phase 1, Quality investigates Phase 2.
• Trending of OOS by product, method, instrument, analyst, time-of-day, reagent lot, standard lot — visible on a Quality dashboard.
• Extension assessment as a structured step that queries batches sharing the suspected root-cause variables.
• Field Alert Report decision automated against 314.81 criteria for stability OOS on distributed batches.
• Audit-trail review of the chromatographic system as part of every OOS investigation — not as a separate annual exercise.
• Effectiveness check on critical OOS investigations at 3, 6, 12 months after closure.

V5 is not a LIMS — chromatographic data acquisition and method execution live in the customer's lab system of record (LabWare, LabVantage, STARLIMS, Empower, OpenLab). V5 integrates with the lab system to receive results and runs the cGMP and investigation workflow around them.

• Sample status flows from the lab system to V5; OOS results auto-open an OOS record with the sample, batch, method, instrument and analyst pre-filled.
• Phase 1 / Phase 2 workflow is a guided form with checklist enforcement — the investigation cannot close with required steps blank.
• Supervisor and Quality Unit involvement is enforced at the right transitions with two-person e-signature where required.
• Extension assessment runs a query against batches sharing root-cause variables (same method, same instrument, same standard lot, same column lot).
• Field Alert Report decision is auto-flagged when 314.81 criteria are met on a distributed batch.
• Trending dashboard surfaces OOS by product, method, instrument, analyst, standard lot, column lot — daily refresh.
• Audit trail captures every state change, every classification, every signature, with reason-for-change on any edit.
• BMR / CoA release is blocked while an OOS investigation is open on a sample required for release.