Quality in Design - Medical Device Design: Innovation from Concept to Market - page 135

mode - in this box you need to describe the imagined failure, in detail. The third column
enables you to write down the effect of this failure mode. The fourth column enables you to
determine the cause of the failure mode. As stated earlier there may be more than one cause
for any failure mode; all have to be examined.
The next four columns enable you to assess the degree of failure. Some failures are
inconsequential; others are really, really bad. Completing this table demonstrates that you
have, at least, thought about them. The first of these ratings columns is “severity” (see Table
7.11(b)) . An example rating is taken direct from the standard associated with risk analysis
(ISO 14971:2009). The worst severity is death (given a 5); the least is a minor inconvenience
(given a 1) - there should never be a zero. Note that the failure of your device may not have
caused the death - your device may have failed and have a knock on effect to something it
interacts with, i.e., a systemic failure, but you must consider this too. However Table 7.11(b)
is only from an injury viewpoint; you can also damage your product's reputation. Consider
the situation where a surgeon is in the OR using your device and it locks up so that it is
unworkable and they have to cancel the operation. Would they ever use your equipment again?
NO! To your company this failure is as bad as a death. As we are dealing with a medical
device it makes sense to relate the effect to the effect on the patient, the person using it, etc.
However we must not forget the effect on the company's “brand.” Hence while death is
clearly a bad outcome, for our company's brand a product recall is equally bad (I know this
sounds crass, but it is a truism). Hence, unlike the later risk analysis, this FMEA covers any
deleterious effect. Hence Table 7.11(c) incorporates severity that also affects your company's
reputation in the marketplace so this table should be used…equally you can develop your own.
The next column in Table 7.11(b) is more problematic as it all depends on how many of your
devices there are and how often they are used. So while I have suggested some guidelines
there may be others; for example you could decide on the occurrence being based on the
number of times a year (if it is a device used infrequently), or a number of failures a month
(if used frequently). For this column just define something and stick to it. One of the most
common ratings is: 1 - rarely; 3 - occasionally; 5 - regularly. The problem is defining what
they mean! The numbers in this column have, again, come from standards (for FMEA it is
BS EN 60812 or its IEC equivalent). This is, again, not suitable for industries that range from
very small to multinational. Hence Table 7.11(d) should be used for this rating.
The last of the three columns in Table 7.11(b) is the easiest to imagine - “detection.” Do
not go overboard with numerous ratings as this will make the analysis difficult later. Stick
to the 5 point scale but set 1 to be detectable by anyone, with 5 being undetectable (except
by someone who has such intimate knowledge - such as you). 3 is in the middle, being
detectable but only by someone who is skilled. You may go for finer ratings if you wish, and
you can refer to other sources, but I have stuck to these just to make our discussions easier.
However, it is not enough to simply fill in the table; we must do something with it. Figure 7.9
illustrates a typical FMEA process.