Keep an OOS investigation from becoming a search for a passing result

A passing retest does not explain the first result

A laboratory obtains an out-of-specification result, then generates passing results during further work. The reporting screen displays only the latest approved number. Reviewers now have a tidy result field but no clear account of the original failure, why additional testing occurred or how all the evidence influenced the decision.

A pharma OOS investigation workflow should make that sequence impossible to overlook. It needs to preserve the initial observation, organise the assessment and control subsequent work. The intended outcome is a scientifically supported conclusion, not a favourable number.

FDA's May 2022 OOS guidance rejects repeated testing to obtain a pass and says that passing retests alone do not justify invalidating an initial OOS result when no laboratory or calculation error has been identified. This is US guidance; the site's approved investigation procedure and applicable requirements govern the actual case. FDA OOS guidance.

Preserve the analytical context immediately

Link the result to the sample, preparation, method version, instrument, analyst, calculation and applicable specification. Keep the original data and relevant metadata accessible. The investigation should not depend on a screenshot of the final value when the analytical record contains more information needed for review.

Record when the result was observed, who was notified and what immediate actions were taken under the procedure. Preserve relevant materials and records according to the approved process. Avoid making up a universal storage interval or disposal rule; suitability for further work depends on the method and sample context.

Separate the initial result from later reviewed conclusions. A result can acquire an investigation status or an authorised interpretation without disappearing from the history. The interface should show that distinction clearly enough that a user cannot mistake the latest displayed value for the complete evidence.

Organise the laboratory assessment around evidence

The assessment should examine plausible laboratory explanations using the actual method and records. This may involve calculations, preparation steps, instrument performance and other relevant evidence. The responsible laboratory specialists determine the appropriate scope; software can organise the work but cannot infer a cause from a checklist score.

Keep proposed explanations identifiable as hypotheses until supported. “Likely dilution error” is not equivalent to a demonstrated event that explains the result. Record what evidence would support or challenge the hypothesis and what the review actually found.

Where the laboratory assessment does not establish an assignable explanation, the workflow must support the broader investigation required by the approved procedure. It should not force users to select a laboratory cause merely because the event originated in QC.

Worked example with an initial failure and later passes

Consider an illustrative assay with an assumed acceptance range of 95.0 to 105.0 percent. The initial reported result is 93.8 percent. Further authorised work produces 99.1 and 99.4 percent. These values are invented to illustrate record handling, not a testing plan or a product-disposition recommendation.

The system should display the initial result and the later results with their sample and preparation relationships. It should also display why the further work was authorised and what question it was intended to answer. A list of three numbers without that context cannot establish whether the work supports a particular explanation.

Simply averaging the values produces approximately 97.43 percent, but that calculation does not explain the original failure. The workflow should not apply such an average as an automatic resolution. The relevant method, investigation evidence and approved data-handling plan determine what calculations are meaningful.

If an investigator proposes a preparation error, the reviewer needs the evidence for that conclusion. If the explanation is not established, the unresolved original result remains part of the evidence considered by the authorised quality function. The application should not turn a selected cause label into permission to suppress the result from the report.

The final record should make the reasoning inspectable: what was observed, what was tested, what was supported, what remained uncertain and who made the resulting decision. Another reviewer should be able to reconstruct that sequence without asking the analyst to recreate it from memory.

Control retesting and resampling as different activities

Further testing should be linked to an approved purpose and the relevant sample identity. Retesting existing sample material and collecting a new sample are different activities with different evidential implications. The application should not present them as interchangeable buttons that create another result field.

Require the appropriate plan, authorisation and expected data handling before the work begins. Preserve any approved amendment and its reason. Do not allow the number of tests to expand informally because the current results are inconvenient.

A useful system control is a visible reconciliation of planned and executed work. If the plan calls for defined activities, the reviewer should see which were completed, which were not and whether any additional work occurred. Missing or extra records need explanation rather than silent exclusion.

Keep the investigation connected to the batch and wider scope

An OOS event may affect more than one sample or batch. The approved investigation determines the scope and relevant manufacturing review. Link affected records so the quality team can see unresolved questions during disposition and subsequent review.

The deviation-investigation guide covers the broader evidence and timeline approach. The LIMS sample-workflow guide explains how testing and review status remain distinct from lot disposition.

Avoid duplicate investigations that reach conflicting conclusions without awareness of each other. If the laboratory and manufacturing systems maintain separate records, preserve the relationship and define which process owns the overall conclusion. Technical integration should support that responsibility rather than create another independent approval path.

Review changes and reprocessing history

Result corrections and analytical reprocessing need context. The reviewer should be able to inspect the original and subsequent records, reasons, responsible persons and applicable approvals. A final report alone may not show the sequence needed to assess the work.

Use the audit-trail review workflow to make relevant changes accessible within the record-review process. The review should follow a risk-based, approved approach and retain evidence of what was assessed. An automatically generated “no issues” label is not a substitute for a meaningful review.

For outsourced testing, agree how complete investigation information reaches the responsible customer quality team. A passing certificate without the relevant investigation context can create a misleading picture even when the laboratory retains the underlying records internally.

Test the workflow with an unresolved case

Use synthetic data to challenge the system with an initial OOS result, later passing results, an unconfirmed hypothesis and an unauthorised attempt to add testing. Verify that the original evidence remains visible and that the next decision goes to the correct role.

Then test an approved correction, a revised investigation plan and a broader affected-batch scope. Check reporting and exports as well as the main screen; a system can preserve the history internally while omitting it from the evidence package reviewers actually use.

NextEdge Labs can help connect the laboratory records, investigation workflow and review package. The first acceptance condition should be that no relevant result is lost when the investigation changes state, and every further test remains traceable to its authorised purpose.

Summary and next actions

An OOS workflow should preserve evidence and support scientific interpretation throughout the investigation. Later passing results are additional evidence, not an automatic explanation or replacement for the original result.

• Review how one recent OOS record links its initial result, source data and subsequent work.
• Define the authorisation and documentation required before further testing begins.
• Test whether reports and exports retain the complete relevant result and investigation history.
• Confirm that qualified reviewers, rather than result-selection logic, control the conclusion and disposition decision.