Two teams scored the same failure mode. One wrote 40 on the FMEA sheet, the other 160.
The failure mode was a leaking gasket on a terminal HEPA housing. The engineering team gave severity 5, occurrence 2, detection 4. A second team, with QA and microbiology in the room, gave 8, 4 and 5. No single score moved by more than three points on a ten-point scale, and the product moved by a factor of four. The scores are invented for the example.
Neither team was careless. The first knew the housing had passed its last leak test. The second knew what had grown on a settle plate in that room.
A risk priority number looks like a measurement because it is a number. It is three judgements on ranked scales, multiplied. The ranks are ordinal, so 160 does not mean four times the risk of 40.
ICH Q9(R1) gave the subject a section of its own. It says subjectivity can be introduced when risk questions are inadequately defined and when tools have poorly designed scoring scales, that it cannot be completely eliminated, and that it may be controlled by addressing bias and assumptions, using the tools properly and maximizing the use of relevant data and sources of knowledge.
On a scoring sheet that comes down to anchoring each scale step to something countable and writing the assumptions beside the score.
What does a 4 for occurrence mean on your site's scale, in events per year?
Topics
Industries




