|
|
|
Risk analyses, including hazard analyses
and various kinds of FMEAs, can be some of the more painful experiences
encountered during the course of medical device development. Done properly,
they are satisfyingly useful tools that can save time and cost in the latter stages
of development. The following tips are based on experience in facilitating
many risk analyses:
-
Don't confuse hazard analyses (HA)
with FMEAs:
HAs are best performed early, during design input development. Most
importantly, they use a top-down approach, that is, the potential
product is evaluated by categories of hazards and not on an individual
part-by-part basis.
FMEAs, on the other hand, cannot be conducted until the design output is
largely finished. They use a bottom-up approach, that is, the product is
evaluated on an individual part-by-part basis.
-
FMEAs identify how parts can fail,
how often they will fail, and the severity of the harm they can cause:
In an FMEA, the effect may be hazardous, or it may just be an annoyance.
In an HA however, hazards are not limited just to part failures, but part
failures that only result in annoyance are not considered at all. Also,
the presence of a hazard does not automatically mean that harm will
result. Therefore, the rate of occurrence of harm will always be equal to or
even less than the rate of occurrence of the hazard.
-
When performing a design FMEA
(dFMEA), you must assume that the device has been manufactured correctly
and contains no manufacturing defects. Conversely, when performing a
process FMEA (pFMEA), you must assume that the device has been designed
as intended and contains no design defects.
-
In an FMEA, be careful to be
consistent in assigning failure modes, causes, and effects:
The failure mode is how a part fails to meet its design intent. This can
mean a part breaking that is not supposed to break, or it could mean a
part wearing out sooner than specified.
The cause listed should be the proximate cause. Sometimes, there are
different root causes. Multiple entries can be made using the form "proximate
cause due to root cause," where the immediate cause of the
observed defect is substituted in place of proximate cause and the
initiating cause is substituted in place of root cause.
The effect must not be confused with the severity of the effect. The
effect is the final manifestation of the failure, such as "incorrect value
displayed" or "artery punctured." It is not the final outcome, such as
death, serious injury, etc., which is the purpose of the severity rating.
-
A problem that sometimes occurs when
assigning severity values is that a single harm or effect might produce
a range of final outcomes that spans two or more severity ranking
levels. The problem is which severity to pick. The answer is to choose
the severity that occurs the most often. If the choice is between two
severities that have the about the same rate of occurrence, pick the
higher severity. Following this approach can save much disagreement
while conducting the risk analysis.
Enter the AlvaMed Website for more information on our
services
 |
