A few months ago in this blog, we discussed the suitability of analytical methods and how a firm needs to demonstrate that a method is fit for purpose and, with that determination, there needs to be an understanding of the measurement uncertainty (MU) associated with the reported results (from that method). This blog can be found here.
Taking this discussion further, one can ask what risk the MU poses to the test data that is generated from a laboratory, understanding that there is MU associated with all generated test data.
The USP-generated article in the Pharmacopeial Forum titled “Fitness for use: Decision rules and target measurement uncertainty” (Burgess, C., et al., January 2016) addresses the risk that the MU has in terms of the material specification and the associated decision rule (DR). The risk that the MU has on the decision rule associated with the specification should also consider how the material specification aligns with the corresponding safety/efficacy of the product.
As per the USP article, “A DR is a documented rule that describes how the MU will be allocated with regard to accepting or rejecting a product according to its specification and the result of measurement.”
The diagram below is from the referenced USP article and it depicts the material specification but introduces the transition zone/guard band concept (here we see an upper and lower transition zone/guard band for a two-sided specification).
Chris Burgess’ article in Pharmaceutical Technology titled “Using the Guard Band to Determine a Risk-Based Specification” (October 2014) calculated the guard band “width” based upon the method variance with a statistical coverage factor. However, it is derived and the statistical approach that is employed needs to be justified.
So, returning to the question, are all passing results equivalent? The answer is clearly “No” as passing results that fall within the transition zone (i.e., those passing results where the error of measurement overlaps with the specification) pose a risk compared to those that fall within the stringent acceptance zone (as depicted in the table above). Results residing in the stringent acceptance zone are associated with a lower level of risk of making an incorrect quality decision where a batch of product is disposition for release when, in fact, it does not meet regulatory specification requirements.
In the transition zone, the firm should recognize the risk and take additional actions, such as initiating a quality investigation to determine the root cause of why the result is in the transition zone and assess the need for additional action. This would commonly be addressed through internal/alert limits that require a deviation to be raised (when those limits are breached). With such a concept, the benefits of driving down a method’s MU are clear as you would want to maximize the “width” of the stringent acceptance zone.
The main message is that it behooves a firm to understand the capability of its analytical methods in terms of error of measurement/measurement uncertainty and then understand the risk that this poses to the associated specification.
If you have any questions related to analytical methods and measurement uncertainty or would like more information on this topic and applicability to your firm, please reach out to us at LCS@lachmanconsultants.com.
