What are the requirements for Medical Device Sampling Plan justification?
I am looking for a guideline on drafting Sampling Plan Justifications.
Sampling plans must be statistically valid. The best way to do this is thru standards such as those available thru ISO and ASQ. But you do need to read the full standard and use it for its intended use as written.
FDA codified this requirement in 21 CFR 820.250. Unfortunately ISO 13485:2016 only speaks in generalities in 8.25 and 8.26 and gives no guidance in “a practical guide” for ISO 13485 either. The sampling standards though May give some help when you learn how to apply them.
I attach two articles addressing this topic. Even though it's old, Torbeck's article should be read before Taylor's.
While there is no definitive guidance on sampling plans from the FDA, have used the following for inspection of incoming components and final product release.
High Risk (from FMEA) use AQL 0.15 or 0.4
Medium Risk AQL 1.5
Low Risk AQL 2.5
Typically, have deployed variable sampling (Z1.9) and double sampling (Z1.4) plans for efficiency.
FMEA Risk is probability of occurrence of Failure and Severity of Effect of Failure.
ISO 14971 Risk is probability of occurrence of Harm and Severity of Harm.
These are quite different and for medical device product safety you must pay attention to these details. The EU MDR and IVDR has codified the ISO 14971?definition of “risk”in those regulations.
FMEA is not a replacement for Risk Analysis, but is a useful tool late in design (after Design Outputs are available) to provide a check that any single-fault failures have not been overlooked. And ISO 14971 does not stop with single-fault, but requires that all failures, not just single-fault, are analyzed.
My point is that all tools should be used within their limitations, and require a complete understanding of the tool for the purpose it is being used. Many companies throw around the term FMEA without understanding the tool and its proper use. And FMEA is NOT Risk Analysis, it is a Reliability tool used for Failure Analysis.