Does anyone have any experience of setting Acceptance Quality Limits (AQL) for medical devices? We are about to launch a new device, it is manufactured and packed by a sub-contractor, we receive and then sterilise and sell. Because it is a new device we would like to inspect the first batches received then reduce the inspection provided there are no problems. The sub-contractor performs their own in-process inspection.
Should we apply an AQL to our own inspection or is this something for our sub-contractor to do and we just spot-check the product?
I'm not great with stats so any advice appreciated!
Thanks,
Philip
Before going into the nuts and bolts of your system, I recommend getting familiar with the basic concepts and terminology, with the aid of overviews such as
this one or
that one. There are many others; I found those by Googling "incoming inspection basics" (1st and 2nd results). As noted, there are also some good discussions on those topics at Elsmar, but I'd go after them only once feeling confident with the base concepts.
By the way, the AQL-based sampling system (such as in Z1.4) is not purely statistical and its development included economical considerations (ie trying to avoid very large sample sizes without sacrificing too much statistical rigor).
I agree that the most important (and the first) step is understanding what risks you are trying to mitigate and what characteristics of the product are to be monitored/controlled. Is it just cleanliness / bioburden? Is it dimensions etc? Is it quantities / packaging configurations? Etc. etc. Each characteristic may lead to different AQLs, different sampling plans etc.
As a general advice, I would mistrust any new supplier and let their actual performance build my confidence over time, modifying my QC practices accordingly. While it's good to have their final inspection records at hand (their in-process inspection would be less relevant), I wouldn't heavily rely on it at first. "Spot check" is just an unfancy name for a sampling plan, with the potential disadvantage of being too loosely defined.