In a message dated 3/24/00 5:07:16 AM Pacific Standard Time,
aris.koumandaros@pharma.Novartis.com writes:
<< Dear Jim Akers,
I certainly agree that overkill is defined as 12D kill of target organisms
with
a D value of 1 minute [Fo = 12 minutes]. As I had stated, use BIs that
provide
good correlation between the physical and biological F values. Hence, for a
validated sterilization process yielding physical Fo values of 24 minutes, it
would make sense, I think, to use BIs that also provide approximately the
same
cycle lethality (such as BIs with a D value of 2 minutes).
Given the BI's spore population and D value from the manufacturer's
certificate
as well as the experimental minimum accumulated F value obtained during the
run,
one may then calculate the "experimental" probability of non-sterility
using the
equation I had indicated in the e-mail.
For example,
Given:
mean population = 1.2 x 10 E+6
D = 2.0 minutes
min. accumulated Fo = 24.5 minutes
Then:
log B = log (1.2 x 10 E+6) - 24.5/2.0
log B = - 6.17
Therefore, B = 6 x 10 E-7.
I would appreciate any comments relating to the calculations.
Best Regards,
Aris Koumandaros
>>
Dear Aris:
The calculation you provided is inappropriate. In order to explain why, let
me start by describing what we are trying to achieve when we use BIs to
determine a good correlation between physically measured Fo and biological
Fo. The following are the steps I'd recommend in sterilization cycle
development.
1. Determine what general sterilization approach you are going to use. By
this I mean bioburden or overkill. (It's important to note that there are
acceptable variations on these to general approachs) For the sake of this
discussion and to be consistent with the model you proposed let's assume an
overkill approach.
2. Establish the lethality requirement that you will target. Depending upon
which overkill model you choose the Fo requirement could be 12-18 minutes.
Note, the process lethality requirement is established without consideration
to BI selection. BI's are used to evaluate the process not to define the
process.
3. Evaluate the commodities being sterilized to determine physical lethality
yield. This is often called a "heat penetration" study. It is extremely
important to make sure that this study is done in such a manner that
thermocouple placement does not create a steam channel, which could have the
effect of making the commodity appear to be easier to sterilize than it
actually is. This study will enable you to determine pre-vacuum requirements
among other things.
4. Once the heat up characteristics of the commodities have been determined
and the cycle run under anticipated production conditions, you should
evaluate the lethality delivered during heat-up and cool down. You'll find
that in a production autoclave the lethality delivered during these phases of
the process is quite consistent. Once the autoclave reaches operating
temperature the lethality delivered per minute will be quite consistent until
the cool-down phase. Therefore, a partial cycle can be run and a BI can be
used to establish that the physical Fo is confirmed biologically under these
reduced cycle conditions. Of course, if there is good correlation between
the BI result and the lethality measured with T/Cs this relationship will be
maintained for the full cycle as well. In practice I target a partial cycle
Fo delivery that will result in about 50% of the BIs surviving. Without
survivors it's impossible to accurately access the process.
5. Once this is done, the typical next step is to run a full cycle test
using total kill analysis. Frankly, this test if of little real practical
value and is done only to satisfy regulatory authorities.
Unfortunately, to keep this post from being exceedingly long, I've had to
simplify this description considerably. This approach is basically the one
taught to me by Dr. Phlug many years ago.
The really important thing I want to stress here is that we as an industry
have taken to using BIs to define our processes when in reality we should be
using them only to evaluate the process. I can evaluate a typical moist heat
process with BIs within a fairly broad range of D values provided I know
precisely what the D value is (and that's another story altogether).
Aris, your problem isn't the equation you are using it's the concept you are
applying. It seems you believe that overkill is a twelve log reduction of
whatever BI you happen to have available for use. This isn't the case.
Overkill is overkill regardless of BI choice. The following is an example of
how an approach such as yours can result in perceived failures. If you have
a BI that has a D val of 2 minutes this means that an 8 log kill will result
in a 0.01% probability of survivors, which equates to a 16 minute cycle. If
you are striving for total kill it's generally wise to consider a ten log
kill as a target that will surely result in no survivors. Note that the
vendore in their C of A will give you both a D value and a survival window.
The survival window tells you the lethality conditions under which survivors
should not occur. However, the D value can vary with respect to media
selection and even the media's contents of specific constituents such a
divalent cations to name but one. Let's say for the sake of discussion that
the actual D value under your test conditions is 2.5 minutes. A ten log kill
of a BI with a D val of 2.5 minutes is 25 minutes. Thus your 24 minute
overkill cycle could result in positives. In this case the problem is
neither the sterilization cycle, or the BI, it's the experimental design.
I frequently here of people saying they have to change their process cycle
because of BIs. This should never happen! An overkill cycle is 12-18 Fo
depending upon which definition you choose (and in my opinion 12 minutes is
adequate considering the safety margin involved). We don't have to change
overkill to accomodate the BI.
Jim Akers
akainckc@aol.com
|