Out of Specification (OOS) Investigation: FDA 21 CFR 211.192 & Root Cause Analysis

Introduction

An Out of Specification (OOS) result is one of those situations in pharmaceutical QC where you need to stop and look at the result properly. Something has gone outside the approved specification, and the first question is usually: Is this a laboratory problem, or is the product actually failing?

After more than 20 years in pharmaceutical QC, I have seen why this question needs a proper investigation. An OOS should not be treated as a number that we keep testing until it comes within specification. The first result is part of the investigation, and we need to understand how that result was generated.

Sometimes the reason is simple. There may be a calculation mistake, sample preparation error, instrument problem, or another laboratory issue. But sometimes the laboratory checks are fine. In that case, we have to look further—at sampling, raw materials, equipment, manufacturing conditions, process parameters, and other related information.

This is where 21 CFR 211.192 becomes important. The regulation requires unexplained discrepancies and failures to meet specifications to be thoroughly investigated. It also requires consideration of other batches or products that may be associated with the failure.

In this article, I will go through the OOS investigation from a practical QC point of view—what I would look at first, when the investigation should move beyond the laboratory, how retesting and resampling should be handled, and how root cause and CAPA should be established.

The objective is simple: find out what happened, understand the possible impact, and take the right action.

Affiliate Disclosure: Some links in this article are affiliate links. As an Amazon Associate I earn from qualifying purchases.

Quick Summary

• OOS means a result is outside an approved specification.
• 21 CFR 211.192 requires unexplained specification failures to be thoroughly investigated.
• Start by reviewing the original laboratory work and data.
• Check the analyst’s work, calculations, sample preparation, instrument, standards, reagents, system suitability, and relevant electronic data.
• If no laboratory cause is found, look at manufacturing, sampling, materials, equipment, and related batches as appropriate.
• Do not keep retesting simply to obtain a passing result.
• Fishbone and 5 Whys are useful tools, but the actual cause still needs evidence.
• CAPA should deal with the problem found during the investigation and should be checked for effectiveness.

What Is an Out of Specification (OOS) Result?

An Out of Specification (OOS) result means that a test result is outside the approved specification or acceptance criterion for a pharmaceutical product, material, component, or applicable in-process test.

This can happen with many different tests.

For example:

• Assay below or above the approved range
• Dissolution failure
• Impurity above its limit
• Total impurities above specification
• Content uniformity failure
• Water content outside the limit
• Microbiological acceptance failure
• Other approved physical or chemical test failures

The investigation depends on the test.

An HPLC assay OOS, for example, requires a detailed look at sample preparation, standard preparation, calculations, chromatograms, system suitability, instrument performance, and other analytical records.

A microbiological OOS will require a different type of review.

The basic question remains the same:

Why did we get this result?

OOS Does Not Automatically Tell Us the Cause

An OOS tells us that a specification was not met. It does not tell us why.

There may be a real product or manufacturing problem. There may also be a laboratory error.

For example, if an analyst used the wrong dilution factor and the investigation clearly proves it, we have a laboratory cause.

But if the calculation is correct, the sample was prepared correctly, the instrument was working properly, and the method was followed, we cannot simply assume that the original result was wrong.

That is why the investigation has to come before the conclusion.

OOS vs OOT

OOS and OOT are not the same.

It means the result is outside the approved specification.

OOT, or Out of Trend, generally means that the result may still be within specification but is unusual compared with previous results or an established trend.

For example, suppose the assay specification is 95.0–105.0%.

A result of 93.8% is OOS.

Now suppose previous batches have normally been around 100–101%, but a new batch gives 95.2%. That result is still within specification. However, depending on the company’s OOT procedure and historical data, it may deserve further review.

So, simply put:

OOS = outside the specification.

OOT = potentially unusual trend, even though the result may still meet specification.

FDA 21 CFR 211.192: What Does It Require?

When dealing with an OOS, 21 CFR 211.192 is one of the main requirements to keep in mind.

The regulation says that an unexplained discrepancy or a failure of a batch or any of its components to meet specifications must be thoroughly investigated, whether or not the batch has already been distributed.

The investigation should also extend to other batches of the same drug product and other drug products that may have been associated with the particular failure or discrepancy. The investigation must be documented, including conclusions and follow-up.

From a QC point of view, the message is simple:

Do not close an OOS just because you think it was a laboratory mistake. Show what happened.

FDA’s OOS guidance gives more practical detail on how these investigations should be handled. The current guidance, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, was revised in May 2022.

How Should an OOS Investigation Start?

The exact procedure should come from the company’s approved OOS SOP, but the general flow is:

OOS result → laboratory review → laboratory cause or no laboratory cause → further investigation where required → root cause and impact assessment → CAPA → final quality decision

I would not start by asking the analyst to repeat the test.

I would first look at the work that produced the original result.

Phase I: Laboratory Investigation

FDA’s OOS guidance describes an initial Phase I laboratory investigation. This is where the laboratory looks closely at the original test and tries to determine whether there is an identifiable laboratory cause.

For an HPLC OOS, for example, I would normally start with the following.

Check the Original Data

Look at the actual data, not only the final result.

Depending on the test, this may include:

• Raw data
• Chromatograms
• Calculations
• Sample preparation records
• Standard preparation
• Sequence information
• System suitability
• Analyst observations
• Instrument information
• Relevant audit-trail data

The point is to understand what actually happened during the test.

Check the Method

Was the approved procedure followed?

Look at things such as:

• Sample weight
• Dilution
• Extraction
• Sonication
• Filtration
• Standard preparation
• Injection conditions
• Calculation formula
• Other method-specific requirements

A small mistake in sample preparation can produce a large difference in the final result.

Check the Instrument

The instrument should also be reviewed.

Depending on the test, check:

• Calibration or qualification status
• Maintenance status
• Error messages
• Instrument performance
• Flow or pressure
• Detector response
• Temperature
• Autosampler performance
• Other relevant parameters

For HPLC, I would also look at the chromatogram itself. Sometimes the chromatogram tells you more than the final calculated result.

Check System Suitability

System suitability should be reviewed against the approved method.

Depending on the method, this may include:

• %RSD
• Retention time
• Tailing
• Theoretical plates
• Resolution
• Response
• Other method-specific requirements

If something went wrong with system suitability, the actual circumstances need to be reviewed. We should not automatically write, “System suitability failed, therefore the OOS is invalid.”

Check Standards and Reagents

Check whether the correct standards and reagents were used and prepared properly.

Look at:

• Standard identity
• Standard status
• Preparation
• Reagent preparation
• Solution storage
• Expiry or retest status where applicable
• Any unusual observations

If a standard or reagent is identified as the cause, the investigation should show how that conclusion was reached.

Check Sample Preparation

This is an area that deserves careful attention.

For many tests, the final result depends heavily on how the sample was weighed, diluted, mixed, extracted, filtered, or transferred.

The calculation can be completely correct while the sample preparation is wrong.

That is why checking only the calculation is not enough.

Check Relevant Audit-Trail Information

For computerized laboratory systems, relevant audit-trail information should be reviewed according to the procedure and the circumstances.

For an HPLC investigation, this may include:

• Manual integration
• Reprocessing
• Sequence changes
• Processing method changes
• Deleted or repeated injections
• Other relevant changes to electronic data

The purpose is not to treat every audit-trail entry as a problem. It is to understand the complete analytical record.

When the OOS involves HPLC testing, the electronic data history can also become important. For a deeper look at this part of the review, see our guide to HPLC Audit Trail Review: ALCOA+ Principles in Pharmaceutical QC.

When Can an OOS Be Invalidated?

An OOS may be invalidated when the investigation establishes a specific and scientifically supported cause for the result.

For example:

• A calculation error is clearly demonstrated.
• The wrong sample dilution was used and the records prove it.
• An instrument malfunction is confirmed.
• The wrong standard or reagent was used.
• Another clear laboratory error is demonstrated.

The important part is the evidence.

A passing retest is not, by itself, proof that the original OOS was caused by laboratory error.

FDA has cited firms for inadequate OOS investigations where results were invalidated without an adequately established laboratory cause.

So I would always ask:

What proves that the original result was caused by this error?

If we cannot answer that, we need to look further.

For readers who want a more detailed resource on laboratory OOS investigations, Investigation of Out-of-Specification Results in the Laboratory by Jerry Lanese is another reference worth considering.

Phase II: Full-Scale OOS Investigation

When the laboratory investigation does not identify a clear cause, FDA’s guidance describes moving into a Phase II full-scale investigation.

Now the investigation may move into manufacturing and other areas.

Depending on the situation, this can include:

• Batch manufacturing records
• Raw materials
• Equipment
• Process parameters
• In-process results
• Sampling
• Manufacturing deviations
• Environmental conditions where relevant
• Previous batch history
• Related batches
• Other relevant quality records

The scope depends on the OOS.

An OOS investigation also shows why QC cannot be viewed as an isolated laboratory function. Product quality is influenced by materials, equipment, manufacturing processes, sampling and quality systems. I have discussed this broader QC approach in Pharmaceutical Quality Control: How Drug Quality Is Built & Tested.

For an assay failure, for example, I would want to know whether there was a problem with blending, sampling, raw material quality, equipment, or a process parameter.

For a dissolution failure, the investigation may need to look at formulation, granulation, compression, coating, raw materials, process conditions, and the laboratory test itself.

There is no benefit in checking everything randomly. Follow the possible causes.

Retesting vs Resampling

These two terms are often mixed up.

Retesting generally means testing the original sample again.

Resampling means taking another sample from the batch or material according to an appropriate sampling procedure.

Both may be appropriate in certain situations, but neither should be used simply because the first result is inconvenient.

The company’s OOS procedure should define how additional testing is handled, and the decision should be scientifically justified.

For example, if the first assay is 92.5% and the next result is 98.0%, we still need to understand why the first result was 92.5%.

The second result does not automatically remove the first one.

Testing Into Compliance

This is one of the biggest problems in OOS investigations.

Imagine:

• First result: 92.5%
• Second result: 94.0%
• Third result: 96.1%
• Fourth result: 97.2%

If we simply choose the passing results and ignore the first result, we have not really investigated the problem.

Why was the first result low?

• Did the sample preparation follow the approved procedure?
• Could sampling have contributed to the result?
• Was there any evidence of product non-uniformity?
• Did the instrument or equipment show any issue?
• Could any part of the manufacturing process have contributed to the OOS result?

Those questions still remain.

FDA’s OOS guidance discusses additional testing and emphasizes that testing should have a sound scientific basis.

The objective should be to understand the OOS—not to keep testing until the result looks good.

Root Cause Analysis

Once the investigation has enough information, we need to identify the cause.

This is where tools such as Fishbone/Ishikawa and 5 Whys can help.

I use these tools as ways of organizing our thinking. They are not proof of a root cause.

Fishbone Analysis

A Fishbone diagram can help the team look at possible causes under areas such as:

• People
• Method
• Machine
• Material
• Measurement
• Environment

For an HPLC OOS, for example, we might consider analyst technique, sample preparation, instrument performance, standard preparation, method requirements, and measurement issues.

But putting these items on a Fishbone does not mean we have found the cause.

We still need evidence.

5 Whys

The 5 Whys method is useful when we want to go deeper than the immediate error.

For example:

Why did the assay fail?

Because the tested sample showed low assay.

Why did the sample show low assay?

Because the sample was not representative of the batch.

Why was the sample not representative?

Because the sampling procedure did not adequately address the product’s distribution.

Now the investigation is moving away from simply saying “low assay” and looking at the process behind the result.

But there is no rule that says we must always ask exactly five questions. Sometimes the cause becomes clear earlier. Sometimes the investigation needs much more work.

Root Cause Is Not Always “Analyst Error”

This is something I would be careful about in any OOS investigation.

If an analyst makes a mistake, we should document it.

But we should also ask why the mistake happened.

• Was the procedure unclear?
• Did the calculation create any difficulty?
• Was the analyst adequately trained?
• Could the worksheet or software have contributed to the error?
• Did the equipment have any issue?
• Has a similar mistake occurred before?

The answer will determine whether the real corrective action is simply analyst retraining or something more.

“Human error” should not become a convenient way of ending an investigation.

CAPA After an OOS

CAPA should match the cause.

If the procedure was unclear, improve it and verify that the change actually works.

For an equipment-related problem, address the equipment issue before closing the investigation.

A sampling problem may require changes to the sampling process.

Training can be appropriate when there is clear evidence that inadequate training contributed to the problem.

But training should not be used as the answer to every OOS.

A good CAPA should have a responsible person, target date, proper documentation, and an effectiveness check where appropriate.

The final question should be:

Did the action actually prevent the problem from coming back?

Batch Impact Assessment

An OOS investigation should not always stop with the batch that produced the result. Where there is a reasonable connection, we also need to consider whether the same problem could have affected other batches or products.

Under 21 CFR 211.192, the investigation should extend to other batches of the same drug product and other drug products that may have been associated with the failure or discrepancy.

Depending on the investigation, the review may include:

• Other batches using the same raw material
• Batches manufactured on the same equipment
• Similar manufacturing conditions
• Previous OOS or OOT results
• Stability data
• Relevant complaints
• Distributed batches
• Other products where a common cause is possible

Equipment-related findings may also require a closer look at cleaning controls, particularly where contamination or carryover could be a possible contributor. In that situation, our guide to Cleaning Validation in Pharmaceutical Manufacturing covers the cleaning process, sampling and analytical controls in more detail.

The review should be based on the evidence from the investigation rather than treating every OOS as a reason to review every batch or product.

A Practical HPLC OOS Example

Let’s take a simple example.

Assume the approved assay specification is 95.0%–105.0%.

The analyst obtains 93.6%.

That is OOS.

I would not simply tell the analyst to repeat the test.

First, I would want to see the original work.

We review the sample preparation, standard preparation, calculations, chromatogram, system suitability, instrument status, and other relevant data.

Suppose we find that the analyst used the wrong dilution factor.

The records clearly show what happened, and the calculation error explains the 93.6% result.

Now there is an identifiable laboratory cause.

The investigation should document it, assess the possible impact on other results, and determine the appropriate corrective action.

Now take a different situation.

Everything in the laboratory looks correct. The calculation is correct. Sample preparation is correct. System suitability passed. The instrument is fine.

A retest gives 98.1%.

Can we simply replace 93.6% with 98.1%?

No.

We need to understand the difference.

If further investigation finds a genuine sampling problem and the evidence supports it, that may explain the original result.

But if no cause can be established, the original OOS remains part of the overall quality assessment.

That is why an OOS investigation is more than repeating the test.

Common OOS Investigation Mistakes

Repeating the test until it passes

This is probably the easiest mistake to make.

A passing result does not automatically explain the failure.

Blaming the analyst too quickly

An analyst can make an error, but the investigation should show that the error actually occurred.

Stopping after the laboratory review

If the laboratory cannot explain the OOS, the investigation may need to move into manufacturing.

Treating Fishbone as the root cause

Fishbone is a tool for looking at possible causes. It is not evidence by itself.

Calling everything “human error”

Sometimes the person made the final mistake, but the system may have allowed that mistake to happen.

Using training as every CAPA

Training is useful when training is actually the problem. It will not fix a poorly designed process or defective equipment.

Forgetting related batches

A failure in one batch may point to a wider issue, which is why the impact assessment matters.

Frequently Asked Questions

What is an OOS result? +

An OOS result is a test result that falls outside an established specification or acceptance criterion.

What should you do when an OOS occurs? +

Document the result and start the investigation according to the approved OOS procedure. Begin with the laboratory work and expand the investigation when the evidence requires it.

Can an OOS result be invalidated? +

Yes, when a specific and scientifically supported cause demonstrates why the original result was invalid. A passing retest alone is not enough.

Can you retest an OOS sample? +

Additional testing may be appropriate when scientifically justified and handled according to the approved procedure. It should not be used simply to obtain a passing result.

What is Phase I of an OOS investigation? +

Phase I is the laboratory investigation described in FDA’s OOS guidance. It focuses on the original test and possible laboratory causes.

What is Phase II? +

Phase II is the full-scale investigation described in FDA’s guidance. It can include manufacturing and other relevant areas when the laboratory investigation does not establish a sufficient cause.

What does 21 CFR 211.192 require? +

It requires unexplained discrepancies and failures to meet specifications to be thoroughly investigated, including consideration of potentially related batches or products, with documented conclusions and follow-up.

Is OOS the same as OOT? +

No. OOS means the result is outside the approved specification. OOT generally refers to an unusual or unexpected trend that may still be within specification.

Bottom Line

An OOS result is not the end of the investigation. It is the point where the investigation starts.

Look at the original work first. Check the method, calculations, sample preparation, instrument, system suitability, standards, reagents, and relevant data.

If you find a real laboratory error, document it properly.

If you do not find one, do not create one just to close the investigation.

Move into manufacturing, sampling, materials, equipment, or related batches when the evidence points in that direction. Then establish the root cause, assess the impact, and put the right CAPA in place.

For me, the most important rule is simple:

Do not investigate an OOS just to get a passing result. Investigate it to understand why the result happened.

References

1. U.S. Food and Drug Administration (FDA). Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production — Guidance for Industry, Level 2 Revision, May 2022. [FDA OOS Guidance]
2. 21 CFR § 211.192 — Production Record Review. [21 CFR 211.192]
3. International Council for Harmonisation (ICH). ICH Q9(R1): Quality Risk Management. This guideline provides principles and examples of quality risk management applicable across pharmaceutical quality activities.
4. International Council for Harmonisation (ICH). ICH Q10: Pharmaceutical Quality System. ICH Q10 provides the quality-system framework relevant to process performance, corrective and preventive action, and continual improvement.
5. FDA Warning Letter — EpiSciences, Inc. Example of FDA concerns regarding inadequate OOS investigation and invalidation of results without adequate supporting investigation.