Dan has provided some great information as always. I would add/clarify that, during your FDA MDR-reportability evaluations when you are assessing whether a malfunction is "likely to" cause or contribute to death or serious injury if the malfunction were to recur, then that absolutely requires a consideration of, and is dictated by, the probability of occurrence of death or serious injury if the malfunction were to recur. That probability rank needs to come from the latest understanding of the subject device's probability of occurrence of death or serious injury (i.e., subtypes of ISO 14971 "harm") from recurrence of the subject malfunction (a subtype of ISO 14971 sequence of events and hazardous situation).
This needs to be paired with proper consideration of FDA's understanding of "likely to". Specifically, when promulgating Part 803 for this context, FDA said that "likely to" means "not remote". FDA has been subsequently pressed on the meaning of "remote / not remote", but has generally deferred and not given a direct answer, yet has emphasized that, whatever way we define these terms, it needs to be defended and documented in the MDR evaluation event file when deciding not to report. That way FDA can later have the opportunity to scrutinize our rationale.
Accordingly, I would suggest that you utilize the/a latest corresponding/relevant state of the art on such matters. For example in particular, you need to correlate these terms to the general probability/frequency categories that you've defined in your ISO 14971 risk management SOP and/or the subject device's Risk Management Plan by which you are ranking your projected/observed probabilities/frequencies of each type of harm.
Remember also that if a particular malfunction of a subject device or malfunction of one similar marked by a firm has ever led to a death or serious injury, then FDA automatically presumes by default that such malfunction is already considered to be "likely to" lead to death or serious injury upon recurrence.
------------------------------
Kevin Randall, ASQ CQA, RAC (U.S., Europe, Canada)
Principal Consultant
Ridgway, CO
United States
© Copyright 2022 by ComplianceAcuity, Inc. All rights reserved.
------------------------------
Original Message:
Sent: 07-Dec-2022 10:14
From: Dan O'Leary
Subject: Probability of Harm versus Likely to Result in Death or Serious Injury
There are two different, but related things going on.
In the pre-market phase, the risk management process evaluates patient or user harms for severity and frequency of occurrence (called probability in the standard) estimates the risk, implements risk control, and evaluates the residual risk for acceptability. The residual risk applies to the device when it ships. It means that you don't expect any complaints that fall outside your estimated severity and frequency of occurrence.
After clearance, the device enters the post-market phase. Complaints come in and follow two different paths.
In one path, evaluate each one for reportability in the MDR system and report (or not) following the criteria in Part 803.
In the other path use the information to revise (or not) the residual risk estimate.
Neither path controls the other. Report regardless of your estimated residual risk. Your estimated residual risk could include reportable events.
Notice that the MDR criteria "likely to cause death or serious injury should the event recur" will be in two different cells in the traditional risk matrix because they have different severities. They may have different frequency of occurrence based on the original estimate and post-market updates.
In the competitor's case, one of two things is happening. Their malfunctions are minor and unlikely to cause death or serious injury were they to recur; they are not reportable. They are not properly applying the reporting criteria and are headed for a Warning Letter.
------------------------------
Dan O'Leary CQA, CQE
Swanzey NH
United States
Original Message:
Sent: 07-Dec-2022 09:24
From: Anonymous Member
Subject: Probability of Harm versus Likely to Result in Death or Serious Injury
This message was posted by a user wishing to remain anonymous
Our company manufacturers a class II device that has been on the market for a long time. FDA's Total Product Life Cycle for the product code indicates industry is filing medical device reports of product problems but no events related to death or injury. Similarly our company has and does file MDRs for product problems which have been assessed as "likely to cause death or serious injury should the event recur"; however, because we don't see deaths or injuries from the complaints we receive, industry is not seeing deaths or injuries and literature searches support the occurrence is low/does not happen, the probability of occurrence of harm is rated as very low in the product's risk assessment documentation. A manufacturer with a similar product recently confirmed they too have the probability of occurrence rated as very low but do file MDRs related to complaints they receive. Can someone explain the difference in thought process between MDR submission and risk assessment probability of harm? Is the "likely" in the MDR submission related solely to severity and not probability of occurrence? I seem to be missing some key understanding.