OOS in dissolution testing: GMP investigation, defensible retest, and inspection-ready justification

Dissolution testing is a test that generates a great deal of data and a great deal of temptation, the worst of which is to repeat the test until it passes. Precisely for this reason, authorities examine investigations under a magnifying glass: a poorly managed dissolution OOS can quickly become a quality system and data integrity problem.

A dissolution result is OOS only if, after completion of the compendial stages, the batch does not meet the criteria (for example, it fails S3). If it fails S1 or S2 but later passes, it is not a final OOS, but it is still an atypical result that should be evaluated and trended. In the case of an OOS: batch hold, Phase I for assignable errors, retest only with documented justification, Phase II with Production and QA, CAPA, and impact extension.

Contents

1. When a failure is truly OOS, and when it is not
2. What to do in the first 2 hours: containment and governance
3. Phase I: assignable errors, technical checklist
4. Retest: when it is allowed and how to make it defensible
5. Phase II: product/process failure, what to really investigate
6. Extension and impact: trends, stability, other batches
7. Effective CAPAs: what convinces inspectors
8. “Audit defense pack”: how to present a case without losing credibility
9. FAQ

1) When is a failure truly OOS, and when is it not

In the dissolution context:

it is OOS when, after the compendial sequence has been completed up to S2/S3 where required, the batch remains non-compliant, for example because the average is below Q or too many units are below the limits.

Be careful, however:

failing S1 does not automatically mean a final OOS, because the procedure requires extension to S2;
but one unit below Q+5 in S1 is still an atypical signal that many companies handle through an investigation record or trending, even if the batch is later found compliant at S2/S3.

Inspection red flag: seeing many batches reach S3 without any system action such as trending or CAPA.

2) First 2 hours: containment and governance

A mature handling approach always starts here:

• batch quarantine / release hold
• immediate notification to QA
• preservation of raw data, with no “creative” reprocessing
• blocking of any unauthorized repeat testing

3) Phase I: assignable errors (technical checklist)

Inspection guidance and the best practices recalled in the document push toward a rigorous Phase I: quick but complete checks to identify evidence, not opinions.

Phase I checklist for dissolution

Data and calculations

• dilution factors, standard potency, formulas, transcriptions
• consistency between vessel/time point/absorbance or HPLC injection

Execution

• was the sampling time respected?
• were medium pH/volume/temperature correct? Was degassing performed if needed?
• were rpm correctly set and actually compliant?
• visual anomalies: floating unit, trapped unit, unit adhering to the vessel, cloudy medium

Filters

• correct filter? possible adsorption? was the first filtered volume discarded if required?

Instrument

• calibration/qualification status
• mechanical checks and temperature, including on that specific day

Golden rule: the conclusion “laboratory error” must be based on facts, logs, and objective evidence. Without that, it is a weak conclusion in an audit.

4) Retest: when it is allowed and how to make it defensible

The guide is very clear on this point: you cannot repeat indiscriminately until a compliant result is obtained; every retest must be justified by an identified potential error and approved according to procedure, while retaining the original result.

A defensible retest = 4 conditions

• a specific error hypothesis, for example incorrect timing, defective filter, instrument anomaly
• evidence, or at least a strong documented indicator
• QA approval before execution
• integrated interpretation: the initial result does not “disappear”; it must be explained in relation to the retest.

5) Phase II: product/process failure (what to really investigate)

If Phase I does not find assignable errors, the guide describes the correct escalation: involve QA and Production to assess process/product causes.

Typical areas:

• API variability, for example particle size distribution
• compression, including hardness, porosity, actual disintegration time
• coating, such as thickness/uniformity for modified release
• process changes, even historical ones, not evaluated for impact on dissolution

Classic investigation mistake: stopping at the first convenient explanation (“it’s the API”) without correlation to good batches or without a full differential analysis.

6) Extension and impact: trends, stability, other batches

Inspectors do not look only at “that batch”:

• they want to know whether it is an outlier or a trend
• they ask for checks on stability, for example decreasing values even if still compliant
• they expect evaluation of related batches, such as same formulation/line/process

Audit-ready best practice: historical graph of % dissolved and frequency of S1/S2/S3, with QA commentary.

7) Effective CAPAs: what convinces inspectors

According to the guide, what “passes” in an audit is not a cosmetic CAPA, but a CAPA with:

• clear link between root cause and action
• effectiveness verification, for example monitoring over n batches
• recurrence prevention, through training, process controls, method update under change control, and RA involvement where needed

8) Audit defense pack: how to present a case without losing credibility

When an inspector opens a dissolution OOS, they typically look for:

• adherence to procedure, with no “informal” retesting
• complete scientific reasoning
• impact/extension assessment
• CAPA and prevention
• link with batch record and release decisions, including QP

Recommended “defense pack” attachments

• event timeline, showing who did what and when
• raw data + audit trail
• signed Phase I checklist
• comparative analysis of good batches versus the OOS batch
• risk assessment if compendial or multi-market differences are involved, using ICH Q9/Q10 as common language

FAQ

If the batch passes at S2 or S3, do I still need to open a deviation?
Many companies do so as an “atypical result” or trending record, because it is a signal of variability and of interest to inspectors, even if it is not a final OOS.

Can I ignore an OOS if “with the other method (USP/EP) it passes”?
No. A failure must be investigated regardless; choosing after the fact the method “that makes it pass” is indefensible and may be interpreted as manipulation.

If you need an operational guide with real examples of failure, S1–S3 criteria, typical errors, checklists, and a risk-based approach able to withstand FDA/EMA scrutiny, you can find “Guide to Pharmacopoeias: Dissolution Testing and Compendial Methods” on guidegxp.com.