Chemistry Reference
In-Depth Information
invalidated result. However, appropriate corrective action should be determined and
implemented to prevent any future occurrences.
6.13.2.2 no Assignable cause/retest
A retest can provide justification that the suspect result should be invalidated; the
investigation may fail to yield an assignable cause, but the result is still considered
to be an aberrant or suspect result. In this case, when the analysis is rerun according
to an approved written retest plan and is successful (result meets the preestablished
acceptance criteria identified in the retest plan), the original results are invalidated
pending acceptable review of the retest plan justification.
6.13.2.3 no Assignable cause/result Is confirmed
Even if the investigation fails to yield an assignable cause, the result may still be
aberrant or suspect. In this instance the analysis is rerun according to the written
retest plan, and the result from reanalysis confirms the original result and must be
considered accurate.
During the laboratory phase of an investigation, a number of practices are used.
These include retesting a portion of the original sample, testing a new specimen from
the collection of a new sample from the batch (resampling), and using outlier testing.
6.13.3 r
eteStIng
Sometimes the investigation may involve retesting a portion of the original sample.
Retesting is often indicated when investigating instrument or sample handling prob-
lems, for example, a suspected dilution error. The retest sample should be taken from
the same homogeneous material that originally produced the OOS result. Decisions
to retest should be based on testing objectives and sound scientific judgment, and
should always be performed by a second analyst (i.e., not the one who originally
obtained the OOS result!). The number of retests should be specified in the SOP,
to avoid “testing into compliance,” or repeated retesting until a passing result is
obtained. If the OOS result is found to be a laboratory error, the retest results are
substituted for the original results. The original results must be archived, however,
and all explanations documented with the proper sign-offs of all involved. Software
that provides for electronic signature sign-off and audit trails helps to maintain
regulatory compliance in this regard. If no laboratory or statistical (mathematical)
errors can be identified, the original OOS results cannot be invalidated, and must be
reported along with the retest results.
6.13.4 r
eSAmPlIng
Resampling is different from retesting because it involves analyzing a new specimen
from the collection of a new sample from the batch, as opposed to the analysis of the
original sample. Resampling is used when it is suspected that the original sample
was not prepared properly, or not representative of the batch. Resampling should
be performed by the same qualified, validated methodology used for the original
sample.
