The outcome of a good Out-of-Specification (OOS) or Out-of-Trend (OOT) investigation should be sufficient data to allow the correct decision regarding batch disposition and the identification of areas for improvement.
OOS/OOT brought about by genuine lapses in attention should be rare. It is more typical that a system failure permitted an individual to initiate a course of action that generated OOS/OOT results. The causes of which may include: unclear or complex processes, poor training, time/resource pressures, etc.
The first steps in an investigation should be a thorough discussion with the individuals involved in the analytical testing. These are the people who have the greatest understanding of how the method performs and often understand subtle differences between test campaigns. Analysts should be encouraged to take note of anything unusual during routine analysis. These observations could critical to the investigation.
The Japanese principle of Gemba or “at the place” is advised to be employed, with the initial discussions taking place in the laboratory. It can be easier to identify potential root causes whilst at the place of analysis as opposed to the office. These observations would fulfil the requirements of a Phase 1a investigation (as defined by MHRA guidance) – Obvious errors. These error types could be highlighted during Point-of-Process checking (second person verification) before results are even generated.
Should Phase 1a not identify the root cause, further review and hypothesis testing can be conducted. Proposed hypotheses should be specific to the investigation and scientifically justified, making sense for the results originally obtained. Hypotheses can be suggested by various Root Cause Analysis tools such as Ishikawa/Fishbone diagrams. Hypothesis testing should include the re-measuring only of original test solutions/articles, not re-testing.
A critical factor to consider is the balance of producing too much data, which can confuse the investigation, and not having sufficient data to draw a conclusion from. For this reason, all hypothesis testing should follow an approved, predefined plan.
Phase 2 investigations should be conducted alongside a manufacturing investigation, for this reason the MAH should be consulted prior to initiation (every effort should be made to involve the releasing QP). Technical Agreements should clearly define this aspect of the investigation where testing is outsourced.
Phase 2 involves the re-testing of original samples, typically multiple times (the number of replicates should be justified) according to a test plan. Whereas Phase 1 investigations should utilise the same analyst that generated the result in question, Phase 2 should utilise a second, equally trained and experienced analyst. If this is not possible, the reason and impact to the investigation is expected to be documented.
Phase 3 involves the resampling of the batch from the manufacturer. This should be conducted in accordance with approved sampling and test plans. In this case, special emphasis should be placed on the justification for accepting or rejecting the original result.
It is important that “cherry picking” of results is not employed during an investigation, true root cause may be hidden if this approach is taken. No data generated as part of the investigation can be used for release purposes. Should a root cause be identified, this should immediately be corrected and a final preparation for release can be created.
The investigation report should chronologically detail all steps taken, including justifications where necessary as well as details of technical discussions. It should conclude with the following; Root Cause Analysis, Risk/Impact assessment and Preventive and Corrective Actions (PACA), evidence of all should be documented. It may be of benefit to assign a defect category or reason to each investigation to aide future trend and effectivity analysis.
There is a cause and a reason for all non-conformances. As an investigator, it is our job to identify these and minimise the risk of them occurring again.