I'd challenge only using P1 if your product is SaMD or if you're calculating risk of your SiMD via SFMEA or other tool. We assume 100%, so P1 is basically not used.
Rhonda Johnston | She/Her | Senior Manager Design Quality Assurance
rhonda.johnston@smith-nephew.com
Original Message:
Sent: 10/5/2023 9:53:00 AM
From: Kevin Randall
Subject: RE: Probability of Occurrence
ISO 14971 / EN ISO 14971 (as amended) Annex C.1 and ISO/TR 24971 sec. 5.5.2 emphasize that the probability of occurrence of harm can be expressed as a combination of separate probabilities (P1, P2) (which can themselves be a combination of sub-probabilities), or can instead be expressed just as a single probability (P). They go on to remind us that a decomposition into P1 and P2 is not mandatory. Therefore, doing so or not doing so depends on which approach makes the most sense for the subject case.
Due to unknown contributing sub-probabilities, I typically focus on the single probability of occurrence of harm (P). And while I acknowledge the scientific rigor required to calculate or estimate true and accurate P1 (i.e., Pa * Pb * Pc * Pd...), I'm also not prepared to declare that this is impossible, nor to declare that 14971 Annex C and 24971 clauses 5.4.7, H.2.7, etc., are thus worthy of dismissal due to these impracticalities.
If the contributing sub-probabilities are accurately known and can be legitimately combined, then by all means they should be factored in, even when there may be incomplete information. Seasoned risk management practitioners understand that a risk profile derived using 14971 risk management is oftentimes no more than an elaborate educated best guess or an elaborated brainstorming output. But oftentimes, it is a practical reality that our best educated guess is the best we can do; and that our best is better than quitting or throwing mud on 14971, and that it is generally good enough for the needs of public health. For example, 24971 says it could be the case that either P1 or P2 may be unknown, in which case a conservative approach can be used by setting the unknown probability equal to 1. I submit that, although we might be able to devise examples where calculating P1 and P2 are impossible, I nonetheless don't believe this means that we must declare it impossible to calculate P1 and P2 for all scenarios.
Regarding that popularly discredited four-letter word FMEA: FMEA and reliability techniques can be used as part of assessing the sequence of events and figuring out P1. Indeed, 14971 says that many sequences of events will only be identified by the systematic use of risk analysis techniques (like FMEA). On that note, it seems to have become quite hip to frown on the use of FMEA as part of risk management. But 14971 and 24971 are clear that, while FMEA alone is certainly not risk management, it remains true that FMEA absolutely, positively, is a valid risk analysis technique, and that risk analysis is a mandatory part of ISO 14971 risk management. Accordingly, I continue to push back on the questionable notion that FMEA isn't a valid part of the risk management process. If we disqualify FMEA or other risk analysis techniques from being allowed as part of 14971 risk management, then we are no longer performing legitimate 14971 risk management.
------------------------------
Kevin Randall, ASQ CQA, RAC (Europe, U.S., Canada)
Principal Consultant
Ridgway, CO
United States
© Copyright 2023 by ComplianceAcuity, Inc. All rights reserved.
------------------------------
Original Message:
Sent: 05-Oct-2023 06:52
From: Richard Vincins
Subject: Probability of Occurrence
Anon,
I have a certain view on all of this, which is probably different than others. There are also quite many posts about risk management in this forum. In practicality, calculating P1 and P2 for medical device would be a long, arduous, and maybe not complete process. The P1 and P2 in the standard is more of a concept view to understand how a hazardous situation and harm resulting from a hazard can be a risk to a user, patient, or person. As an example, if you take a typical customer complaint it would be quite difficult to gather specific, detailed, or finite information related to the hazardous situation. In these cases, medical device companies get a complaint that the product stopped working or the patient was injured. Further investigation might not reveal anything specific. So for me, probability of occurrence (P) is the one to use for medical devices. Electronics industry, automotive, aerospace, might be a bit easier because they can attribute the failure of a device to specific actions or activities, especially if software driven. Finding P1 for non-electronic products, OTC products, or products like catheters, yea, that would be challenging activity.
Edit: Just saw the other post and response by Dan, would refer to that also - good information.
------------------------------
Richard Vincins ASQ-CQA, MTOPRA, RAC
Vice President Global Regulatory Affairs
Original Message:
Sent: 04-Oct-2023 03:54
From: Anonymous Member
Subject: Probability of Occurrence
This message was posted by a user wishing to remain anonymous
What is the consensus on the use of separate probabilities (P1, P2) for expressing the probability of occurrence of harm (ISO 14971:2019 Annex C) or just a single probability (P).
I believe that the electronics industry has already moved in this direction but conventionally the medical device industry has used one (P) in the use and design FMEAs, but is there a move to separate probabilities, (P1) for the probability of a hazardous situation occurring and (P2) probability of a hazardous situation leading to harm?
Has anyone in the medical device industry moved to the use of separate probabilities and if so what has been your experience with the regulators – how have they perceived it?