Cpk ≥ 1.33 is written into many validation reports as an acceptance criterion. The GMP texts those reports cite do not contain it.
The index compares the distance from the process mean to the nearer specification limit with three standard deviations. At 1.33 that distance is 4 standard deviations. The figure is a convention from manufacturing statistics.
What the texts do say is less specific. EU GMP Annex 15 (2015 text) says statistical tools should be used, where appropriate, to support conclusions on the variability and capability of a process during ongoing process verification. The FDA process validation guidance of 2011 recommends that a statistician, or a person trained in statistical process control, develops the methods used to measure process stability and capability. Neither names an index or a value.
The weakness of the number is the data under it. A Cpk assumes a process in statistical control and roughly normal data. With 10 batches, the lower 95% confidence bound of an observed 1.33 is about 0.8, so the true capability may be below 1. Thirty batches bring the bound back to about 1.
We would keep 1.33 as a target and stop using it as a pass mark on small data sets. The report says more when the control chart, the batch count and the interval sit beside the index.
How many batches stand behind the Cpk in your last product review?
Industries




