OOS vs OOT Results: How to Investigate in Pharma QC

An out-of-specification (OOS) result is a test result that falls outside the approved specification or acceptance criteria. An out-of-trend (OOT) result is still within specification but departs from the pattern expected from previous data, such as earlier stability time points or historical batches.

Both need investigation, but the rules and urgency differ. OOS investigations are a regulatory expectation with a defined structure, and weak OOS handling is a recurring theme in inspection findings and regulatory warning letters. OOT detection is your early warning that a process or product is changing before it fails.

OOS, OOT and atypical results compared

The regulatory framework

The key reference is the US FDA guidance Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, first issued in 2006 and revised in 2022. Although written for the US, it is widely used as the benchmark by Indian and GCC manufacturers, especially those exporting. The MHRA has also published guidance on OOS and OOT investigations. Check the current versions of both before updating your SOP.

The FDA guidance applies to chemistry-based laboratory tests on finished products, APIs, components and in-process materials, including stability testing. Microbiological and biological tests follow the same principles, though some details differ.

Phase I: the laboratory investigation

Phase I asks one question: is there an obvious, documentable laboratory error? It should start immediately, before solutions, standards and glassware are discarded.

The analyst should report the result promptly to the supervisor and preserve everything used in the test. The supervisor then makes an objective assessment, typically using a checklist:

• correct method version and specification used
• calculations and data processing, including chromatogram integration
• system suitability and instrument calibration status
• reference standards, reagents and solutions within expiry and correctly prepared
• sample identity, preparation steps and dilutions
• analyst training and any unusual observations during the test

If a clear laboratory error is identified and documented, the original result is invalidated and the test is repeated. Assigning a cause without evidence, for example blaming analyst error by default, is a classic inspection finding.

Phase II: the full-scale investigation

If Phase I finds no assignable laboratory cause, the investigation expands. Phase II includes a review of manufacturing, involving production and QA, alongside any further laboratory work.

Manufacturing review

Review the batch record, deviations, equipment events, raw material lots and in-process results. Check whether other batches made with the same materials or equipment are affected. The aim is to find out whether the result reflects a genuine product problem.

Retesting, resampling and averaging

• Retesting is testing a portion of the same sample. The number of retests should be fixed in your SOP before testing starts, not decided by the results, and preferably done by a second analyst.
• Resampling means taking a new sample from the batch. It is justified only when the original sample is shown to be unrepresentative or compromised.
• Averaging can hide variability. Report all individual results, and do not average an OOS value with passing results to produce a passing figure.
• Outlier tests: the FDA guidance cautions against using them to reject results from validated chemical tests.

Repeated testing until a passing result appears, known as testing into compliance, is unacceptable. All results, including the original OOS, remain part of the record and the batch decision must consider them. Strong ALCOA data integrity is what makes this traceable.

Detecting OOT results

OOT detection only works if you define expected behaviour in advance. Common approaches include:

• Control charts for release data, such as Shewhart charts with defined rules for shifts and runs.
• Regression-based limits for stability data, comparing each new point with the prediction interval from the batch's own earlier points or from historical batches.
• By-time-point comparison, comparing a result with the distribution of historical results at the same time point.

Whatever approach you choose, write the rules into your SOP so that analysts are not making subjective judgements.

Worked example: an assay OOS on a stability sample in Baddi

A tablet manufacturer in Baddi has a product with an assay specification of 95.0-105.0% of label claim. The 12-month long-term stability result for one batch is 93.8%. Earlier results were 99.6%, 99.1%, 98.7% and 98.4%.

1. Phase I: the analyst informs the supervisor and keeps the solutions. Calculations, integration, system suitability and standard preparation all check out. Re-injecting the retained solution gives 93.9%, ruling out an injection problem.
2. Phase II starts: QA opens a full investigation. The manufacturing review finds no deviations, and other batches on the stability programme show normal trends.
3. Hypothesis testing: the investigator notices the analyst used a nylon syringe filter, while the validated method specifies PVDF. A documented experiment on fresh preparations from the same sample shows the nylon filter retains about 5% of the active.
4. Conclusion: the root cause is a laboratory error supported by experimental evidence. The original result is invalidated, and the predefined retest using PVDF filters gives results consistent with the trend.
5. CAPA: filter type is added to the worksheet as a check, analysts are retrained, and QC reviews other recent tests where nylon filters may have been used.

Note what made the invalidation defensible: a specific hypothesis, a planned experiment and data, not simply a better retest result. Had no cause been found, the confirmed OOS would have needed stability and regulatory impact assessment through deviation and CAPA processes.

Common mistakes

• Discarding solutions before the supervisor has reviewed them.
• Concluding analyst error without evidence.
• Retesting until a passing result appears.
• Resampling without proving the original sample was unrepresentative.
• Averaging an OOS result with passing results.
• Having no defined OOT rules, so drifting stability data is noticed only when it fails.
• Closing the investigation without checking other affected batches.

Key takeaway: An OOS result stands until evidence proves otherwise. Investigate in defined phases, follow retest rules set before testing, and use OOT rules to catch problems before they become failures.

Frequently Asked Questions

What is the difference between OOS and OOT?

An OOS result is outside the approved specification, so the batch cannot be released while it stands. An OOT result is within specification but departs from the expected pattern, such as a sudden change between stability points. OOS investigations follow a regulatory framework, while OOT handling is defined mainly by your own SOP and trending rules.

How many retests are allowed in an OOS investigation?

There is no universal number. The FDA guidance expects the number of retests to be specified in advance in a written procedure, based on sound scientific principles, rather than decided after seeing results. Retests should preferably be performed by a different analyst. Confirm the approach in your SOP and with your QA function before testing begins.

What is a Phase I OOS investigation?

Phase I is the initial laboratory investigation to find out whether an obvious, documentable laboratory error caused the result. The analyst and supervisor review calculations, instrument performance, standards, reagents, sample preparation and the method used. If a clear error is found, the result can be invalidated. If not, the investigation moves to Phase II.

Can an OOS result be invalidated?

Yes, but only when an assignable cause is identified and supported by evidence, such as a calculation error, an instrument failure or an experiment confirming a sample preparation fault. Invalidation without evidence is a common regulatory finding. All original data must be retained and the reasons documented in the investigation report.

Is an OOT result a deviation?

It depends on your quality system. Many sites handle OOT results through a dedicated trending or laboratory investigation procedure rather than as a deviation. However, if the investigation finds a process or product cause, a deviation, CAPA or change control may follow. Define the route clearly in your SOPs so OOT results are never left without an owner.

Next steps

Our Root Cause Analysis & CAPA course builds the investigation skills behind defensible OOS closures, and Data Integrity & Computerised System Validation covers the data controls inspectors test during OOS reviews. For related reading, see deviation vs change control vs CAPA and ALCOA data integrity explained. You can also speak to a counsellor.