Sampling for stability studies - Class II surgically invasive devices

[northernscientist]

Hi

We make Class II surgically invasive devices and I'm trying to find some official guidance on what would be an acceptable sample number for a new stability trial. The tests are generally going to be pass/fail. I know that AQL sampling isn't really appropriate for this kind of thing but I don't know where to look to figure out how to determine the sample size numbers scientifically. I should mention my statistics knowledge is a little rusty, so I might need something that's a pretty basic guide.
In the past I've been challenged by FDA on choice of sample numbers and it was very difficult to answer!
Can anyone help?

 

[Ronen E]

Hi

We make Class II surgically invasive devices and I'm trying to find some official guidance on what would be an acceptable sample number for a new stability trial. The tests are generally going to be pass/fail. I know that AQL sampling isn't really appropriate for this kind of thing but I don't know where to look to figure out how to determine the sample size numbers scientifically. I should mention my statistics knowledge is a little rusty, so I might need something that's a pretty basic guide.
In the past I've been challenged by FDA on choice of sample numbers and it was very difficult to answer!
Can anyone help?

Hello northernscientist and welcome to the Cove

I don’t think that the FDA has generic guidance for that. If there are clearly defined expectations for your device type I would look for them in guidance officially referenced on the FDA website, in the relevant 3-letter code’s page.

From general acquaintance with the FDA and their expectations I would say you’ll need at least n=30, but it’s difficult to say when I don’t know what the device is, what’s the intended use etc. If you could somehow translate the test from attribute (pass/fail) to variable (numeric value) it is likely to reduce the required sample size (however I wouldn’t expect to go below about n=10 for statistical rigor).

If you don’t get useful responses in this thread in a week or so (it takes some time for people to actually see a question, especially during the weekend), you may repost this question in the statistical methods forum - they might not be able to tell you about FDA’s guidance/expectations, but may be able to provide sample size guidance from a purely statistical perspective.

Cheers,
Ronen.