OOS Investigation in Pharma: Complete Guide (FDA, MHRA, EU GMP)

In short: An out-of-specification (OOS) result is any test result outside the limit that applies to the material or product. It must be investigated, in writing, before anything is decided about the batch. This guide walks QC analysts, QA officers and students through the laboratory phase, the full-scale phase, retesting, averaging, outlier tests, invalidation, batch decision and CAPA, and compares the FDA and MHRA approaches.

What is an OOS result?

FDA’s guidance uses “OOS” for every test result that falls outside the specifications or acceptance criteria set in a drug application, a drug master file, an official compendium or by the manufacturer. In-process laboratory tests are included. It covers testing of active ingredients, excipients and other components, in-process materials and finished products, including testing against stability specifications.

Three terms are often confused:

An OOS result triggers a formal investigation every time. An OOT or atypical result needs a proportionate look, and the pattern may reveal a problem before a batch actually fails. The detailed OOT treatment is in Out-of-Trend results: meaning, detection and MHRA expectations.

Regulatory basis

You do not need to memorise clause numbers, but you should know where the expectations come from.

In India, Schedule M and the CDSCO inspection practice expect the same discipline, and most firms that supply regulated markets follow the FDA, MHRA or EU approach. Your own SOP and your quality agreements decide which one is binding on your site.

The investigation at a glance

1. The analyst sees a suspect or OOS result and keeps all test preparations.
2. Phase I: laboratory investigation by the analyst and supervisor. Is there a clear, documented laboratory error?
3. If a laboratory error is clearly proven, the result is invalidated and the test is repeated as the SOP allows.
4. If not, Phase II: full-scale investigation led by the quality unit, covering production, sampling and any extra laboratory work defined in advance.
5. The quality unit evaluates all results and decides: confirmed OOS, invalidated, or inconclusive.
6. Batch disposition, impact on other batches, regulatory reporting where applicable, and CAPA.

Phase I: the laboratory investigation

The aim is to decide whether the OOS result reflects a problem in the measurement or in the product. FDA asks that this initial assessment be done, whenever possible, before the test preparations are discarded, so that hypotheses (a dilution error, an instrument fault) can be tested on the same solutions.

What the analyst does

• Checks the result against the specification before discarding anything.
• Keeps test and standard preparations if they are stable, and tells the supervisor.
• Documents any obvious error straight away, for example a spilled sample or an incomplete transfer.
• Does not carry on with an analysis already known to be invalid just to see what result comes out.

What the supervisor does

FDA lists the points the supervisor’s assessment should cover. In plain terms:

• Talk to the analyst and confirm the correct procedure was followed.
• Examine the raw data, chromatograms and spectra for anything unusual.
• Check that calculations are correct and that no unauthorised or unvalidated change was made to an automated calculation method.
• Confirm instrument performance, and that standards, solvents and reagents were suitable.
• Compare method performance with validation and historical data.
• Document the whole assessment.

Only a clear, evidence-based laboratory cause justifies invalidation. FDA is explicit that an OOS result should not be put down to analytical error without an investigation that establishes a laboratory root cause. “Analyst error” written as a conclusion, with no evidence, is not enough.

For the full walkthrough see Phase I laboratory investigation (US FDA).

Phase Ia and Ib in MHRA terms

MHRA splits the laboratory stage in two. Phase Ia is a check for a clear, obvious error, such as a power failure, an equipment failure, a calculation error or a spilled sample, which is documented and the analysis repeated. Phase Ib is an investigation by the analyst and supervisor, using a laboratory investigation checklist, when no obvious error is found; re-measurement of the original preparations may start only after a hypothesis plan is documented. If neither finds an assignable cause, the case moves to Phase II. Details are in OOS Phase Ia and Ib (MHRA).

Phase II: full-scale investigation

When Phase I does not show a laboratory error and the results appear accurate, the quality unit runs a full-scale investigation under a predefined procedure. FDA expects it to be given the highest priority. It should include:

• A review of production records and sampling procedures, involving every department that could be implicated (and every site, if a contract manufacturer or several sites are involved).
• A check whether the same problem has occurred before.
• A written record: the reason for the investigation, the manufacturing aspects that may have contributed, the documentation review with an actual or probable cause, the history check and the corrective actions taken.
• An evaluation of the effect on batches already distributed.

If this work confirms the OOS result and finds the root cause, the product is rejected, but the investigation must still extend to other batches and products that could be affected.

See Phase II full-scale investigation (US FDA) and OOS Phase II (MHRA).

Retesting and resampling

Retesting means analysing more portions of the original, homogeneous sample material. It is used, for example, to investigate an instrument malfunction or a suspected sample handling problem. FDA’s key points:

• The number of retests is fixed in advance in a written SOP, based on sound science and the variability of the method. It must not change according to the results obtained.
• The SOP should define when testing stops and the batch is evaluated.
• The plan often includes a second analyst who is at least as experienced as the first.
• Departures from the SOP should be rare and need a protocol approved by the quality unit before extra testing starts.
• If no laboratory or calculation error is found in the first test, there is no scientific basis for replacing the original OOS result with passing retest results. All results, passing and failing, are reported and considered.

FDA does not give a universal number of retests. The number belongs in your SOP and must be justified.

Resampling means analysing additional units from the original sampling or a new sample from the batch. It follows predefined sampling procedures. It can show that the first sample was prepared improperly or was not representative, but if the sampling method itself was inadequate, a new method must be developed, documented and approved by the quality unit.

Repeating tests until something passes is “testing into compliance”, which FDA calls unscientific and objectionable. The background is in the Barr Laboratories case.

Averaging and outlier tests

Averaging

Averaging is legitimate when the test method itself defines it, for example replicate injections from the same preparation that give one reportable result, with acceptance limits for variability written in the method. It is not appropriate to average the original OOS result with retest or resample results from the investigation, because this hides variability.

FDA gives an illustration: with a 90 to 110 percent limit, results of 89, 90 and 91 percent average 90 percent, and the retests tend to confirm the original low result. Results of 80, 85 and 105 percent also average 90 percent but show high variability and unreliable data. In both cases the individual results, not the average, should be evaluated. FDA also advises caution where some individual assay results from the same sample are OOS and some pass, all within the known variability of the method: the average is treated as OOS even if it is within specification.

Outlier tests

An outlier test is a statistical tool, not a root-cause finder. FDA says that for validated chemical tests such as an assay of a homogeneous composite, an outlier test must not be used to invalidate the suspect result. It may be used only for information on how far a result is from the rest. Outlier tests have no place in content uniformity, dissolution or release rate, where variability is the thing being measured. If an outlier test is allowed for any application, the SOP must name the test and its parameters in advance.

Invalidating a result

Per FDA, invalidation of a discrete result may be done only upon the observation and documentation of a test event that can reasonably be determined to have caused the OOS result. A theory without evidence is not enough. MHRA’s Inspectorate has written that the deficiency it most often sees is an OOS result invalidated on assumptions and theories with no real evidence behind them. Where you do invalidate, record the cause, the evidence, who reviewed it, and what was done to stop it happening again.

Concluding the investigation and batch decision

The quality unit interprets the results and decides. FDA describes three outcomes:

Release after an inconclusive investigation is possible only when a full investigation shows that the OOS result does not reflect batch quality, and FDA tells the quality unit to err on the side of caution. Also consider low-but-passing results: they may indicate a batch that could go out of specification before expiry.

For applications approved by FDA, an OOS on an already distributed batch can trigger a Field Alert Report within three working days, unless the result is found invalid in that time (21 CFR 314.81(b)(1)(ii), as explained in the guidance). In the EU, §6.35 of Chapter 6 (on-going stability) asks that confirmed OOS results affecting batches released on the market be reported to the competent authorities.

CAPA and trending

An investigation is not finished with a batch decision. FDA’s guidance points out that corrective and preventive action is implicit in the investigation requirement. Typical actions are method or training fixes, equipment maintenance, a change in a process control, or a revision to the SOP. Laboratory error should be rare; repeated “analyst errors” indicate a system problem such as poor training or badly maintained instruments. Management should watch trends in OOS and OOT results as part of the quality system. See CAPA process step by step for how to plan and verify actions.

Records of the investigation, raw data and all results, including the invalidated ones, are retained with the batch records: at least 1 year after the expiry date of the respective product batch (EU GMP Part I Chapter 4 §4.11, or 5 years after QP certification if longer; 21 CFR 211.180(a)), and longer if local law or the marketing authorisation requires.

FDA and MHRA terminology compared

The two approaches are consistent in spirit: prove a laboratory cause before invalidating, decide the retest plan in advance, and never test into compliance.

Mistakes inspectors commonly raise

These come from the FDA guidance and the MHRA Inspectorate blog, not from a warning-letter count:

• Retesting without a predefined plan, or changing the plan after seeing results (testing into compliance).
• Invalidating an OOS result on an assumption such as “probably analyst error” or “possible pipetting mistake” with no evidence.
• Discarding test preparations before the laboratory assessment is finished.
• Averaging the OOS with passing retests to produce a compliant number.
• Using an outlier test to remove an inconvenient assay value.
• Stopping the investigation at the laboratory when the cause might lie in manufacturing, or failing to check other batches.
• Missing or changed raw data. Data integrity gaps make any OOS conclusion unreliable; see data integrity problems as the main topic of FDA warning letters.
• No CAPA, or CAPA that only says “retrain the analyst”.

Read the detailed guides

Frequently asked questions

How many retests are allowed?

Neither FDA nor this guide gives a single number. FDA says the maximum number of retests must be specified in advance in a written SOP, based on sound scientific principles and the variability of the method, and must not be adjusted according to results. Your SOP must say when testing stops and the batch is evaluated.

Can an OOS result be invalidated?

Yes, but only when a documented test event that can reasonably be determined to have caused the result is found, for example a proven dilution error or a failed system suitability. Suspicion alone does not qualify. If there is no clear laboratory error, the investigation continues into Phase II.

What is the difference between OOS and OOT?

OOS is a result outside the specification limit. OOT is a result that is unusual compared with earlier data, though normally still within limit. EU GMP Chapter 6 expects both to be investigated, and your OOT procedure should describe how trends are detected.

Can I average the OOS result with passing retests?

Not as a way of showing compliance. FDA says averaging the original OOS with retest or resample results hides variability. Evaluate the individual results.

Can an outlier test be used to reject an OOS assay result?

FDA says no for validated chemical assays of a homogeneous sample: the test does not identify a cause, so it cannot invalidate the result. It may only be used for information.

Can a batch be released after one passing retest?

Not on that basis alone. If the first test had no identified laboratory error, a passing retest does not replace the OOS result. The quality unit must consider all results and complete the investigation.

Does the FDA guidance apply to Indian companies?

It is non-binding guidance, but it applies in practice to firms supplying the US market, and Indian regulators and customers’ auditors expect equivalent control. Follow the requirements stated in your own quality system and quality agreements.

References

1. US FDA, Guidance for Industry: Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, Revision 1, May 2022 (CDER). fda.gov/media/158416/download.
2. US FDA, 21 CFR 211.165, Testing and release for distribution. ecfr.gov.
3. US FDA, 21 CFR 211.192, Production record review. ecfr.gov.
4. US FDA, 21 CFR 211.180(a), General requirements for records. ecfr.gov.
5. MHRA, Out-of-specification investigations, GOV.UK, published 28 August 2013, updated 26 February 2018; attached presentation “Out of Specification & Out of Trend Investigations”, October 2017. gov.uk/government/publications/out-of-specification-investigations.
6. MHRA Inspectorate blog, “Out of Specification Guidance”, 2 March 2018. mhrainspectorate.blog.gov.uk.
7. European Commission, EudraLex Volume 4, EU GMP Part I Chapter 6: Quality Control (revision in operation from 1 October 2014), §§6.7, 6.9, 6.35.
8. European Commission, EudraLex Volume 4, EU GMP Part I Chapter 4: Documentation, §4.11.