[Aaron]

Joe,

Expanding on what Niki wrote, this is an important topic that seems to lead to a lot of confusion and flawed methodology. The P1, P2 probabilities are part of the ISO 14971 risk model but that doesn’t mean they should be explicit columns in a risk table. My recommendation is to make sure that everyone involved in risk assessments understands the P(harm) = P1 x P2 model but that only P(harm) is listed in the risk docs.

I’ve worked on many Risk Management Files under different quality systems over the past 20 years and I’ve seen a lot of mistakes in how P1 and P2 are combined. Simply taking the minimum of P1 and P2 as the P(harm) value is not going to be accurate in all cases. Theoretically it should be a multiplication of the P1 and P2 probability values. Unfortunately, this is not practical because most companies use rankings (e.g., 1 – 5) instead of actual probability values and you can’t multiply the rankings. Some companies use a complex lookup table to derive P(harm) from P1 and P2 rankings but those have to be very carefully designed to not distort the combined probabilities across all ranges. That’s why I think it’s more robust to list only the overall P(harm) in the released risk documents.

There’s a related question here, too, about whether FMEAs should even include Probability of Harm or should be limited to only Probability of Failure but that’s a topic for another thread.

-Aaron