How to Investigate an OOS Result
Handle an out-of-specification result without testing into compliance.
What a how-to is not
A how-to is SPEQ’s practitioner method, not a procedure. It does not replace your own SOP, it is not a validated approach, and the judgement calls in it belong to your quality unit.
An out-of-specification (OOS) result is a data point until proven otherwise. The investigation decides whether it reflects a laboratory problem or a real quality signal about the batch — and the two failures that sink it (invalidating a result without cause, and "testing into compliance") are exactly what the Barr ruling and FDA’s OOS guidance were written to prevent.
- 1
Report it — never quietly retest first
Report the OOS to your supervisor and open a formal investigation before doing anything else. Retesting in the hope the number goes away is testing into compliance, and it destroys the credibility of the record.
- 2
Run the Phase I laboratory investigation
Review the analytical records for a documented, assignable laboratory cause — a mis-prepared standard, a dilution error, a mis-set method. A result can only be invalidated on a cause you can point to, never on an assumption.
- 3
Use hypothesis testing where Phase I is inconclusive
If no obvious cause is found, a limited, pre-approved hypothesis test may check one specific theory. What makes it legitimate is that the plan and acceptance criteria exist before the data — not after.
- 4
Escalate to a Phase II full-scale investigation
If no assignable laboratory cause is established, the OOS may reflect the product, not the test. Widen the investigation to manufacturing and to other batches that could share the cause.
- 5
Disposition honestly
With no assignable cause, treat the OOS as a real result — reject the batch, or make a fully documented science-based disposition, and make any required notifications. Never average the OOS with passing results to mask it.
- !Testing into compliance — retesting until a passing result appears, then releasing.
- !Invalidating the original result without a documented assignable cause (the central Barr violation).
- !Averaging a failing result with passing ones to bring the mean in-spec.
- !Closing at "must have been lab error" without ever looking at manufacturing.
How to Investigate an OOS Result: frequently asked questions
Common questions on investigate an oos result.
Can I retest an OOS result?
Only under a pre-defined, scientifically sound plan, and never as a first move to make the number go away. Unbounded retesting until a passing result appears is "testing into compliance" — the practice the Barr ruling prohibited. Retesting follows the investigation; it does not replace it.
When can an OOS result be invalidated?
Only when the Phase I laboratory investigation identifies a documented, assignable cause — a specific analytical error you can point to. Invalidating a result on an assumption ("must have been lab error") without that evidence is the central violation the OOS guidance and the Barr ruling address.
What is the difference between Phase I and Phase II?
Phase I is the laboratory investigation — did something in the lab cause the result? Phase II is the full-scale investigation that follows when no laboratory cause is found — it widens to manufacturing and to other potentially affected batches, treating the OOS as a possible real quality signal.
Can I average an OOS result with passing results?
No. Averaging a failing result together with passing ones to bring the reportable value in-spec masks a real signal about the batch and is expressly prohibited. Each result stands on its own within the investigation.