Page 60 - Safety Risk Management for Medical Devices

Usability Engineering and Risk Analysis 39

8.8 USABILITY AND RISK
Some Use Failures can result in Hazards. The UMFMEA analyzes all Use Failures,
and captures the Hazards that are due to Use Failures in the End Effects column. Just
as in other FMEAs, the End Effects which are Hazards are then captured in the Risk
Assessment and Control Table for risk estimation.
Risk is the product of P 1 , the probability of occurrence of a Hazardous Situation
and P 2 , the probability of experiencing Harm from the Hazardous Situation. P 1 itself
is the product of P(Hazard) and P(Exposure to the Hazard). Ordinarily, P(Hazard) can
be derived from the Occurrence rating in FMEAs. However, with respect to risk,
IEC 62366 [14] Annex A, Subclause 5.5 makes the following statements:
"Selection of the HAZARD-RELATED USE SCENARIOS can be based on the SEVERITY of the
potential consequences of the associated HAZARDS. It can be needed in this way to focus on
HAZARDS rather than RISKS because the probability of occurrence of encountering a HAZARD,
which is one component of RISK, can be very difficult to estimate, especially for a novel
MEDICAL DEVICE for which no POST-PRODUCTION data are available.

Another basis for selection of the HAZARD-RELATED USE SCENARIOS is the RISK of the occurrence
of HARM to the PATIENT or USER. These values can also be difficult to determine, as they are
based on assumptions closely related to probability of occurrence and without data, can be diffi-
cult to justify. Finally, and only in the presence of data that provides a justification, should RISK
values based on the combination of SEVERITY and probability of occurrence of the HAZARD be
used as the basis for prioritization of HAZARD-RELATED USE SCENARIOS. Values for these proba-
bilities or probability of occurrence can be derived from POST-PRODUCTION data on current or
previous versions of the same MEDICAL DEVICE or on the level of certainty that the RISK
CONTROL measures are effective, which should also be justified with data."
In other words, the Standard [14] suggests that unless you have data, you should
presume the probability of exposure to Hazards due to Use Failures is 1, and judge
the use scenarios based on the severity of associated Harms.
ISO 14971 [3,7] provides some guidance in section D.3.2.3 that when estimate of
P 1 cannot be made on the basis of accurate and reliable data, or when a reasonable
qualitative estimate is not possible it is best to set P 1 5 1 and focus on using Risk
Control measures to eliminate the Hazard, or reduce the probability of Harm, or
reduce the severity of the Harm.
Naturally, elimination of a Hazard is best, but not always possible. The manufac-
turer has two options:

1. Gather data, upon which P 1 could be estimated, and thus risk could be
computed
2. Assume P 1 5 1, and use Risk Controls to reduce the probability occurrence of
Use Failure as far as possible, and add controls to reduce the severity of the Harm
In the following subsections, each option is explored.