01Name the material used in the follow-up

FDA’s OOS guidance distinguishes retesting original homogeneous sample material from resampling additional units or a newly collected batch sample. That pharmaceutical definition helps prevent an ambiguous request for another test from concealing a change of material.FDA — Investigating Out-of-Specification Test Results for Pharmaceutical Production (opens in a new tab)

The table describes levels of follow-up, not a mandatory sequence. The appropriate choice depends on the proposed cause and on whether suitable original material remains available.

Use sample and preparation identifiers so readers can distinguish these actions in the final report. The word duplicate alone does not establish what was duplicated.

02Match the action to the suspected problem

Consider an original example in which an unexpectedly low result may reflect an incorrect preparation volume. Reinjecting the same solution leaves that volume error in place. It can examine injection behaviour, but it does not independently reproduce the original preparation.

A new preparation from appropriate original material can address a different question. It still needs a documented comparison: how were the amounts and volumes recorded, and does the evidence support the proposed error?

If the concern is that the original selected vial was atypical, collecting other units adds information about the lot. Those units do not reveal the original vial’s true content automatically. Both a low original unit and typical later units could be real.

03A later pass does not invalidate the earlier result

FDA warns against repeated testing to obtain a passing outcome and states that passing retests do not provide a scientific basis for invalidating an initial OOS result when no causative error is identified.FDA — Investigating Out-of-Specification Test Results for Pharmaceutical Production (opens in a new tab)

For an original case, a low result from unit A and a passing result from unit B may indicate unit variation, analytical variation or both. Calling B the replacement result hides the fact that two different units were measured.

If a documented error is found, explain how it affected A and which corrected or additional result addresses that error. If no cause is found, preserve the disagreement as part of the evidence.

Selecting the best result, averaging until the number falls within a limit or stopping only when a preferred outcome appears changes the decision process. Those choices need a scientific rationale established independently of the desired result.

04Make the investigation reproducible on paper

A clear plan names the hypothesis, material available, proposed comparisons, number of follow-ups and how all outcomes will be interpreted. It also identifies what the work cannot establish if the original sample has been consumed or changed.

The final account should link each value to its unit, preparation and method. Record why additional material was selected and whether the original selection procedure itself was questioned.

That level of description allows another reader to judge whether the investigation narrowed the possible causes. The goal is a supported explanation, even when that explanation remains incomplete, rather than a larger pile of unconnected test numbers.