Is GMP Internal Auditing a Sampling? – Part I

We’ve often said that GMP Internal Auditing is a sampling. We don’t Audit everything, only bits and pieces. The question that is often asked, is what size should those pieces be? How many records should we examine? How many people do we ask a particular question to? How long do we observe a particular action?

All of these questions center around selecting the appropriate sample size for GMP Internal Auditing purposes.

But before we look at sample sizes, let us gain a basic understanding of what a sample is. Let’s say you are going to Audit Purchasing to determine if you are using approved Vendors. You might have 5000 individual purchase orders that would apply to the Audit. This would be called the “population”, the total number of possible Audit records (in this case). It might not make sense to check every single record. So, instead, you might want to Audit just a part of the 5000. That would be a “sample”. Now the question is how many of the 5000 records are we going to look at?

First of all, cGMP does not give us a specific sample size to consider. In fact, cGMP does not even hint that Auditing is a sampling. So, what does one use to determine what size of sample to use when Auditing?

  • Statistical Validity
  • Time
  • Complexity
  • Criticality

Statistical Validity

There are formulas available which will lead you to determine what would be a statistically valid sample for a given population. In GMP Internal auditing, there is no requirement to have a statistically valid sample size. Sample size will, typically, be based on the other factors, rather than statistical validity, and that’s okay.

…To Be Continued