When reviewing health agency citations, investigators are naturally concerned when observing cases of data invalidation during investigations and may well question the motives/justification accordingly. This can lead to some uncomfortable exchanges between firms and health agencies!

So how should a laboratory protect themselves from such a situation? First, you want to ensure that, per your quality system, data invalidation can only occur via a thorough investigation, where the Quality Unit approves the justification for such invalidation. This is critical, as awkward a conversation may be when an investigator questions the rationale within an investigation for invalidation; the level of discomfort will increase exponentially if the investigator uncovers cases of data invalidation which were not associated with an investigation and where there was no quality oversight. Such a situation could have far-reaching regulatory repercussions and may lead to significant agency actions. Therefore, you must also ensure robust quality oversight of all data generated by the laboratory, such that any such cases of repeat testing are detected as part of laboratory data disposition.

When dealing with investigations that address data invalidation, the Quality Unit review must recognize the risks and challenge the rationale/justification for data invalidations. A consideration to evaluate such rationale through the lens of an agency investigator where one automatically assumes an ulterior motive. One needs scientific rationale supported with data to challenge such a preconceived notion where the data demonstrates the relationship between cause and the impact to the reported result. Hypothesis testing must occur when dealing with a probable cause and the Quality Unit must recognize the risk of invalidating data based on a probable cause and thus insist on a hypothesis test plan. When dealing with OOS investigations with only a probable laboratory root cause, a Phase II manufacturing investigation must be insisted upon by the Quality Unit and all data must be reported and considered with the batch disposition decision.

What are the other considerations with investigations associated with data invalidation? The Quality Unit must recognize that, when there is a justified root cause for data invalidation, the focus then turns to risk/impact to historical data. The investigation must ask the question of potential impact to historical data. Further, does the root cause impact other test methods? Remember – just because the data met material specification, protocol acceptance criteria, etc., does not mean the data has not been impacted by such a laboratory cause. The investigation must address the risk of false positives and negatives and thus the risk of detrimental impact to material on the market. The outcome of such risk assessments will drive the scope for the impact assessment (governed by a protocol). The Quality Unit should approve the rationale for the repeat testing, be it repeat release testing or repeating the testing associated with a protocol. Such rationale should center on the need for a corrective action to address the root cause (where direct data has demonstrated effectiveness of the corrective action) and where there is confirmation that the repeat test data is not impacted by the root cause. It is understood that it is tempting to say that favorable data from the repeat testing (i.e., data meeting specification/criteria) is no longer impacted by the laboratory root cause, but the Quality Unit must recognize such a risk and require robust evidence to demonstrate accordingly (linking back to the data/evidence that enabled confirmation of cause).

When considering the risk to other test activities / other methods, the investigation must ask the challenging question of how such a test method / laboratory process was established and approved without an awareness of the root cause and why it took such an incident to identify the deficiency within the laboratory method process. Essentially, does this incident indicate a deficiency with a laboratory system(s) such as method development, method validation/transfer, method lifecycle controls, reference standard program etc.? This leads to another consideration, that of trending. Trending needs to occur as part of the investigation, but also outside of the investigation where the laboratory monitors and has metrics relating to the number of cases of data invalidation and that these are periodically evaluated at the site quality council level. It must be recognized that data invalidation should be infrequent and closely monitored as repeat cases may indicate deficiencies at the laboratory system level necessitating associated corrective actions along with expansive retrospective impact assessment along with immediate additional level of oversight (as an interim action). As you can imagine, repeat cases of data invalidation is likely to prompt the investigator to question the laboratory controls along with the effectiveness of the associated systems and thus the validity of all generated data.

If you have any questions relating to how to address data invalidation, then please reach out to Paul Mason at P.Mason@LachmanConsultants.com.