Page 100 - Safety Risk Management for Medical Devices

Risk Analysis Techniques 79

Also, consider potential Hazards that could be encountered under reasonably fore-
seeable misuse conditions. See Section 3.1.1.1 for a definition of “reasonably foresee-
able misuse.”
Interfaces are a common source of failures. Pay particular attention to interfaces to
parts of the Systems that are designed by an external entity (supplier). Remember to
consider potential Hazards arising from interfaces between use-conditions. For exam-
ple, the maintenance function is not commonly in the forefront of the mind of medi-
cal device designers. Could a maintenance person leave the device in a potentially
unsafe state for the clinical user?
Postulate the potential Hazardous Situations for the System, and perform an FTA
to determine pathways that could lead to the identified Hazardous Situations (see
Section 12.1 for instructions on how to perform an FTA).
Populate a RACT template with information from the FTA. All applicable
Hazards from the CHL should appear in the RACT. Each Hazard will have one or
more Hazardous Situations associated with it. Document the potential pathways to
each Hazardous Situation. These pathways are easily derived from the FTA. Fill in a
P 1 value for each pathway. P 1 is the probability of occurrence of the Hazardous
Situation. If field data is available for P 1 , use that. Otherwise, using the collective
team judgment estimate a value for P 1 . When estimating P 1 , use a basis that makes
sense for your product, and facilitates your Risk Control decision making. For a long-
term implantable product, a unit such as patient-year is suitable. But for a device that is
used repeatedly, per-use makes more sense. For example, P 1 for the Hazardous
Situation of “Over-infusion of insulin” due to an erroneous blood glucose reading
could be stated as “10 23 per use.”

Tip Estimate P 1 for one device, not the entire fleet of devices. The reason is that making more
devices doesn’t make any single device less safe. If P 1 is estimated over the whole fleet of devices,
the risk for a given Hazard would appear to continuously grow as more copies of the device are
sold.

The next step is to evaluate the Hazardous Situations and identify the potential
Harms that could ensue from each of them. In the BXM method, for every Harm
there are five P 2 numbers, which are the probabilities of sustaining Harm in different
severity classes (see Section 11.6 for more information on the HAL). Look up the P 2
probabilities in the HAL and populate the RACT table with their respective values. If
your method uses a single P 2 number, then just use that.
Compute the risks for each Hazardous Situation by multiplying the P 1 for each
Hazardous Situation by the five P 2 numbers from the Harms. This will result in five
risk-numbers—one for each severity class.
Compute the residual risks for each Hazardous Situation, and also for the overall
System (see Section 17.3 for details on how to do the computation).