### Why spend 100 hours measuring and testing if there is enough evidence after 15 minutes?

When completing a process validation it is of course important to have sufficient evidence to support the conclusion.

Evidence is usually expressed as Cpk and/or Ppk values and the sufficiency of the evidence is backed by the term confidence.

Confidence as a *general* term is built by use of proper methods, trained staff, SOP’s etc. and can be hard to put a number and still; no matter how high, it is not exactly *evidence* since there are no measurements to back it up.

The process validation must provide *objective* evidence and that is why the statistical term confidence interval (CI) is used. The CI is a calculation that tells you what different result you *might* get if you tried to repeat your experiment (e.g. re-run an OQ).

If the worst value (High or Low) in the CI meets the requirements then You can be X% confident that the requirements are met – that is a way to provide *objective* evidence for Your quality.

However You must remember that *You* decide **how confident You need to be**

.

CI depends on sample size and on variation. To put this in other words: If You take a large sample You will be more certain. If there is a small variation You can settle for a smaller sample size.

Deciding on sample size is important as a large sample will make You more confident but it will require more resources. The cost of the parts may be low but the man hours spent measuring and waiting for results can be costly and can be a bottle neck in a project.

So why spend 100 hours measuring and testing if there is enough evidence after 15 minutes?

You can see from the graph that the Cpk estimate converges to 4.2 after just 15 samples, so why spend twice as much time to reach the same value from 30 samples?

And why not consider repeated testing with smaller samples to speed the investigation up?

Good luck with Your sampling and validation……stay tuned for more