Last month we considered the elements of CAPA. This month we will be more “philosophical” and consider two things relevant to a CAPA investigation. Each item discussed is part of making sure we have identified the correct root cause. One is related to the investigation and corrective action proposed, and the second is related to confirmation that the corrective actions actually were effective. Each is discussed in a separate section below.
A. Cause vs. Correlation
One of the key activities of a CAPA investigation is to determine the root cause. While a “fishbone” analysis may help identify possible causes, the ideal is to determine the actual, true or real cause. In that way changes can be proposed. It is important not to jump to conclusions without consideration of all possible causes. The root cause has to be one that could likely produce the observed effect (the effect is the deviation or the nonconformance observed). Part of a FMEA is to determine the likelihood of a suspected root cause of actually leading to the observed deviation/nonconformance. For example, if the nonconformance is only observed in the filling stage of a liquid drug product, it is not likely that the root cause was related to anything having to do with an earlier process step such as the blending step, although it could be related to the transfer process from blending to filling (as well as to something involved at the filling stage).
For cleaning validation nonconformances, there may be clue as to the root cause based on the nature of the observed nonconformances. For example, if the equipment is swabbed, do high residue values occur in all sampled location or in one or a few swabbed locations? If in limited swab locations, the nature of the location, particularly if it involved manual cleaning and particularly if it was were determined to be a more difficult to reach location in the cleaning process, those facts may suggest a possible root. Even in that example, it is important not to immediately identify the root cause the operator (or operator compliance with the cleaning process SOP). While it may be the case that operator training is the actual root cause, a further and more critical cause might be the lack of adequate detail in the cleaning process SOP itself.
In trying to identify the cause, it might be assumed that just because two things happened at the same time that one caused the other. For example, if there were an upset in a Purified Water system such that high microbial levels were found in the water system monitoring, such as upset could be more likely root cause if the nonconformance in the cleaning process also involved high microbial counts. If the nonconformance in the cleaning process were merely high results in the measured API levels, then it is not likely that the water system upset would be the root cause of the cleaning nonconformance. However, such an event may cause us to further investigate the water system upset and see if there were other factors (beside high water microbial levels) that could have affected the cleaning process and led to the nonconformance.
While a correlation of two things happening close together may lead us to investigate a cause and effect relationship, the mere fact that they happened at the same time may be a coincidence rather than a causal relationship. Or, it may actually be that a third event actually caused the two correlated events, or that when we thought A was the cause of B, it was actually the reverse in that B was the cause of A.
These considerations are part of the reason that regulatory agencies will emphasize the need for thoroughness in a root cause investigations of a deviation/nonconformance. I should also point out that regulatory agencies simultaneously ask that such investigations be timely, and there may have to be a balance between the thoroughness and the timeliness so we actually have an acceptable outcome.
B. Evidence vs. Intention
This issue is more related to what we do after identifying the root cause and establishing a corrective action. The issue here is that we have selected a proposed change to improve the cleaning process. We do this change because we believe the change will likely reduce the re-occurrence of the deviation/nonconformance. That is, we intend that implementation of the change will make a significant improvement. That designed change may be based on common sense, sound scientific judgment, and/or on laboratory or commercial scale studies. However, the true test of whether we identified the correct root cause and then implemented the corrective action should be based on evidence that the deviation/nonconformance is not happening again. That is why we do what is often called an “effectiveness check”.
For a validated cleaning process, the evidence may be something we see immediately, or it may be something that takes a longer time to establish that our corrective action was appropriate. To establish that evidences of success, we may have to increase the level of monitoring data that is routinely collected. For example, if the nonconformance only involves a certain swab location, then perhaps the routine monitoring data may only involve collecting swab analytical data at that one sampling location. On the other hand, if the nonconformance involves a visually dirty surface, then we should still continue with a robust visual examination, but should be paying more attention to trending data and responding quickly to any future similar non-conformances. Such a confirmation of effectiveness is needed to truly close out a CAPA.
I will also add an additional (and more philosophical) concern for CAPA investigations. That concern is the difference between knowledge and wisdom. Knowledge generally means we know what is true (or know what is a fact). Wisdom is more related to what we do with what we know. In the context of this Cleaning Memo, wisdom is needed in order to decide what, and to what extent, corrective/preventive actions should be done based on what we found to be true (or reasonably true) in the investigation.
Copyright © 2023 by Cleaning Validation Technologies