Ï "N10" <limbic_lesion at hotmail.com> Ýãñáøå óôï ìÞíõìá
news:cpqe1m$ole$1 at sparta.btinternet.com...
>> "The Sceptical Chymist" <polyzoni at otenet.invalid> wrote in message
> news:cpq9j4$dp$1 at usenet.otenet.gr...> > Hello everybody.
> >
> > I would appreciate any comments on the following:
> >
> > We are a company that makes frozen dough products, some of which include
> > cured meat in the filling. We're having a problem with one of our
> > customers,
> > a major food multinational, and I would like to hear the opinions of
> > others.
> > The situation is as follows:
> > Two of our products contain boiled ham and bacon, which are placed in an
> > open dough base (they're exposed). The cooking instructions state
clearly
> > that the product should be baked at 180 deg C for15-20 min. However the
> > customer believes that the end-user might let the product thaw rather
than
> > take it straight from the freezer to the oven and additionally might
pick
> > and eat the ham or bacon. For that reason they treat those two
ingredients
> > as ready-to-eat and ask for an "absence of Salmonella spp in 375g"
> > analysis.
> > We're using a Biomerieux Minividas instrument, and the protocol asks
for:
> >
> > 1. Homogenising 25g of sample in 225 ml of buffered peptone water (BPP).
> > 2. Incubating the resulting 250ml of homogenate at 37 °C for 16-20 hrs.
> > 3. Subculturing 0.1 ml of homogenate in 10 ml of Rappaport Vassiliadis
> > (RV)
> > broth.
> > 4. Incubating the RV broth at 41.5 °C for 6-8 hrs.
> > 5. Subculturing of 1 ml from the RV broth in 10 ml of M-broth.
> > 6. Incubating the M broth for 16-20 hrs at 41.5 °C.
> > 7. Heating 1 ml of M broth at 100 °C for 15 min.
> > 8. Adding 0.5 ml of heated M broth to the Vidas SLM test strip and
> > reading
> > the results in 45 minutes.
> >
> > Positive results are followed by confirmation by a reference method that
> > involves plating the selective broth and running biochemical (API
strips)
> > and/or serological (agglutination) tests.
> >
> > If we were to follow the protocol to the letter we would have to test 15
> > 25g
> > samples (the instruments can only run 12 samples at a time) with all the
> > reagents, incubator space, staff and time requirements that would
entail.
> > What we do instead is pool the 15 samples of 25g to 4 samples of 100g
> > (that
> > makes 400 rather than 375g). We dilute those 1:10 with BPP and
homogenise,
> > ending up with four homogenates of 1000 ml. These we incubate at 37 °C
for
> > 16-20 hrs and then draw 0.1ml and subculture it in RV broth. From that
> > point
> > onwards we follow the rest of the protocol to the letter.
> > The probability of the 400g sample carrying Salmonella remains the same
as
> > long as the sample is representative and drawn from different points.
What
> > does change is the probability that 0.1 ml drawn from 1000 ml of
incubated
> > homogenate carry Salmonella. There's a better chance of detecting it if
> > 0.1ml is drawn from 250 ml as the protocol requires. However we believe
> > that
> > after incubating the homogenate at 37 °C for 16-20 hrs in BPP, if
> > Salmonella
> > is present it will multiply to such levels that this difference is
> > insignificant.
> > I'm interested in other people's opinion on the matter, so I would like
to
> > hear from you.
> >
> > I will be posting this to more than one food list/newsgroup, so
apologies
> > to
> > anyone who reads it more than once.
> >
> > Sincerely
> >
> > Kostas Polyzonis
> > QA Manager
> > Michael Arabatzis AVEE - Hellenic Dough
> >
> > To reply by private e-mail type replace invalid with gr
> >
> >
>> Hello Micheal
Hi Des
I'm Kostas, Michael is my boss :) (Michael Arabatzis AVEE - Hellenic Dough
is the name of the firm I work for)
> I agree that sampling plan and proceedure for bulking samples is fine
but
> how was the mass of 375gram calculated is it in itself even significant
as
> a statiscal proportion of your production mass ?. You should know that
375
> gram is not a magical figure. The key factor in deciding how much sample
to
> examine depends on the total mass produced and the limit of sensitivity
at
> which you decide to test. Working to a even 95 % confidence limit is
> relativel trare in the Food Industry .
>> In my experience the majority of Salmonella sampling plans fall far short
> of even the most simple statiscal model. I think you should review the
maths
> to make certain your sample makes statistical sense otherwise all other
> considerations or questions of effciencies are less relevant.
I've already pointed out to that customer, who asked for this particular
sampling plan, that 375g is arbitrary if we don't consider the size of the
complete lot. 375g for one ton of ingredient and 375g for 10 tons? I think
not. They seem stuck on it and we're trying to please them or reach some
compromise
> Assuming you are satisfied with sampling plan then please also consider
the
> following ;
>> A) Large BP volumes may go extremely anaerobic during pre-enrichment
> therefore with volumes of 1000 ml it is advised stir or shake.
Continuous shake during incubation, a good shake before sticking it in the
incubator, or what?
Isn't Salmonella a facultative anaerobe that can survive both aerobic and
anaerobic conditions?
> b) If your dough is "live" you may have extremly high levels of
competative
> organisms during the temeprature equilibriation stage of the BP pre
> enrichment and possibly during some stages of the subsequent incubation.
I was refering to testing two of the ingredients of the filling (ham and
bacon) not the final product.
> Addtionally you going to need an understanding the effects of negative pH
> swing during preencrichment . Do you know if the buffering capaicty of the
> broth is sufficient...check it out I think you might be very supprised !
That shoulds be easy to check.
> I would recommed therefore that you conduct validation trials to ensure
> that low levels of Salmonella ( 1-5 cells per 25 50 or 75 gram) can be
> enriched satisfactorily. I d reccomend working with atleast 5 or 6
strains
> taken from each of the most common O antigenic ( B,D,C,E,F,G).
I'm a little out of my debth here, but how does one innoculate a certain
weight of sample with a certain number of cells to check recovery?
> It may be difficult but I recommend you try to make some assesment of
> quantative levels of Salmonella obtained in the BP broth after pre
> enrichement. In my experience a decent pre enrichment will yeild at least
> log 2 cells per ml after the 16- 20 hours incubation. Theoretically
> selective enrichment will work on log 1 but more of that latter but if you
> do the figures you'll see LOG 2 is attainable post pre enrichment even
with
> stressed or fragile Salmonella populations.
Meaning a hundred times the initial count, right?
> Quantation at this stage could easily be performed by surface platting on
> XLD or BG (etc) and I know this because I ve done it. Given known
> pre-enrichment yields from known levels of intial inoculation you will
then
> be in a position to calculate whether or not your 0.1 ml aliquot fromthe
> 1000 ml culture would contain Salmonella at what ever CI you choose.
C: Confidence? I?
> Factually no one can tell you with defendable accuracy that your 0.1 ml
> aliquote is sufficient... I suspect it is but I wouldnt put my
professional
> name to such a statment rather I truelly beleive you need to validate
your
> method and thus generate your own confidence.
I think validation is the key here, both to convince the customer I was
talking about and to be confident ourselves that our results are
meaningfull.
> You may care to check yields ( I consider it vital) at the end of
selective
> enrichment and the M broth stage. You may care to check the yield with M
> Borth incubated at 37'c instead of 41.c 'c again you might be suprised.
>> C) If you do the work and validate your pre enrichment stage I think you
> still have a problem and its to do with the RV broth. RV is a superb borth
> but single broth selective enrichment for Salmonella species invovles a
risk
> of producing False negative results ( ie missing true positives) .
>> In the case of RV broth you have two factors which need consideration.
> Firstly it is well documented in literature that numerous common
Salmonella
> serotypes are Methyleneblue sensitive, temperature sensitive mutatants are
> also relatively common. We isolate 3-4 methylene blue Salmonella from
food
> samples on a weekly basis and perhapes 2 -3 temperature sensitive mutants
> per month ( Usually grooup D of C1).
>> RV contains methylene blue and is a high temperature incubation ( you use
> 41.5'c)
>> My point is R.V. is " good" but as advised in every international standard
I
> ve read pertaining to Salmoenlla isolation it is always true that at
> least two selective enrichment broths are used to give the highest
> statistical chance of recovery and susequent detection.,,,irrespective of
> the end confirmation proceedure.
>> You are obviously interested in obtaining the "correct result " so I
mention
> these points because no matter what the PR for Vidas system states it is
> flawed if implies that a single broth selective enrichment stage is
> superior or even equivalent to a dual broth system in the long view...
again
> I speak from experience on this one .
>> Best regards Des akaN10
Biomerieux has already validated Minividas. If we depart from their
prescribed method and, for example, start running two different selective
broth silmutaneously (isn't that what you're suggesting?) we might as well
go back to a "slow" method altogether: Plate the selective broths on two
different solid media, look for typical colonies, pick colonies and plate
again for cleaning up, check Gram smears under microscope, triple sugar
iron, and so on.... And after a copuple of weeks we''ll have the results and
tell production they can go ahead and use that bacon or ham. We invested in
the Minividas to avoid exactly that scenario.
Thanks a lot for the input. That's exactly the sort of discussion I was
hopping for when I posted here.
Regards
Kostas
--
Kostas Polyzonis
QA Manager
Michael Arabatzis AVEE - Hellenic Dough
To reply by private e-mail type replace invalid with gr
