30 replies to this topic

[wayne]

Dear All

I am not making attempts to re-open the wound but out of interest, can someone tell me:

1.What is the microbiological criteria / sampling plan that the Cadbury has established for their chocolate?
2.What is the set value of c in their sampling plan?
3.What is the likelihood of obtaining 'false negative' test results in some of their MPN test method?
4. What is the statistical inherent error(s) in the test method?

Thanks & Regards
Wayne

[Simon]

I do not believe this important topic has been raised on the forum before, so I've split it from the Cadbury's debate to form a separate thread.

The issue is also much wider than the Cadbury's recall.

Thanks Wayne.

Regards,
Simon

[Charles.C]

Dear Wayne,
It seems likely that current interest would be on a detect / non-detect procedure.
The typical 2class plan, (using a c =0 criterion) at its most stringent (ICMSF) requires (60x25g) samples / lot.
The 'false negative' is I guess (not a statistician) equivalent to the plan's characteristic which states that with 95% probability, application of such a plan would reject 19 out of 20 lots containing 2 Salmonellae cells per kg. Increased strictness will require a larger unit sample size and / or more samples. ( another viewpoint is that a -ve detection result gives an approx. 95% probability that the lot contains not more than 5% of contaminated units).

If you want to know what a -ve MPN result means, can try

http://www.cfsan.fda...bam/bam-a2.html

note - above link not working, an approximate replacement although perhaps condensed is -
http://www.fda.gov/F...M/ucm109656.htm
(added 19-01-2010 plus the "quoted"paragraphs below)

Mathematically, zero positives has posed a difficulty in that the MPN equation is not specifically solvable. Hence comments as in above link -
'the outcome with all negative tubes is listed as less than the lowest MPN for an outcome with at least one positive tube.'

Unfortunately, this option has it's own theoretical / practical problems , eg

Prior to this revision, the 8th edition tables showed the MPN for the (0,0,0) outcomes as less than the MPN of the (1,0,0) outcome. This made good numerical sense, but made for unacceptable complexity in trying to write acceptance standards for raw materials in terms of the BAM. This revision returns to the prior practice of recording the MPN for the (0,0,0) outcomes as less than the MPN for the (0,0,1) outcome, so that standards can once again be written in a simple manner in terms of all-negative outcomes.

The above is reasonably understandable, the following reference paragraph is much less so, to me anyway, but probably is irrelevant for practical users. i enclose it just to demonstrate the complexity.

Since no particular density is indicated for an outcome of (0.0.0), a density must be assigned arbitrarily (and stated explicitly in the report) in order to calculate statistics. For the logarithm of the density, log[0.5*MPN(1,0,0)] is a reasonable choice. For statistics using the (non-logarithmic) density itself, calculate once with a density of 0.0 and once with a density of 0.5*MPN(1,0,0). Either report both statistics or report one statistic accompanied by a comment on the difference between that statistic and the other one.

http://intra.fb.uner...ium/Chapt10.htm

All this typically leads to the usual indicator expression for decision-making problems, eg -

"The 95% CI reflects the range of uncertainty as usual"

Rgds / Charles.C

[wayne]

Dear Wayne,
It seems likely that current interest would be on a detect / non-detect procedure.
The typical 2class plan, (using a c =0 criterion) at its most stringent (ICMSF) requires (60x25g) samples / lot.
The 'false negative' is I guess (not a statistician) equivalent to the plan's characteristic which states that with 95% probability, application of such a plan would reject 19 out of 20 lots containing 2 Salmonellae cells per kg. Increased strictness will require a larger unit sample size and / or more samples. ( another viewpoint is that a -ve detection result gives an approx. 95% probability that the lot contains not more than 5% of contaminated units).

If you want to know what a -ve MPN result means, can try

http://www.cfsan.fda...bam/bam-a2.html

Mathematically, zero positives poses a difficulty in solving the MPN equation hence comment in above -
'the outcome with all negative tubes is listed as less than the lowest MPN for an outcome with at least one positive tube.'
The 95% CI reflects the range of uncertainty as usual.
Rgds / Charles.C

Dear Charles C.

It is very kind of you in responding to my enquiries. Thanks.

After having read the FDA article, my engineering friend ( a transformed FSMS ISO22k auditor from QMS) is taking a step back and wonder whether he is still qualified in carrying out an ISO22K audit. Certainly, he is an honest & responsible person.

The 2 class plan (c=0), as what the ICMSF has documented (recommended?), needs 60x25g samples/lot will certainly be sending many (if not most) food companies to the wall. Yet, it only gives a 95% CI. Under a food safety policy, it is commonly set at zero tolerance (i.e. 100% adherence) and NOT 6 sigma. The 5% uncertainty means 5 % of the consumer will have to be poisoned (correct me if my calculation is wrong).

MPN does consider the presence of interference in a food sample. But, with the presence of a very low number of pathogen, would the MPN method be able to pick it up under a lowest dilution with the presence of preservative(s) and other unknown inhibitor(s)? Again, what's about those injured cells? Can they be all resuscitated under this defined system? If the validity of the analytical method cannot be confirmed, under the ISO22K requirements, then the control measure is in doubt. Then, the food safety control plan has to be reviewed or be thrown out of the window; alternatively, double the food price or shut down the food business .

I am not sure what is the validity (and sensitivity) of the MPN method being employed under the 2 class plan? Should ( can) we enrich the sample prior to testing? After all, it is all about presence / absence. In your experience, what is the probability of giving a 'false positive' in this system?

Regards
Wayne

[Charles Chew]

Dear Charles C. / Wayne,

I am responding to the title of this forum as one of the most important (if not the most important issue) currently confronting food safety today.

In the light of so much uncertainties on the issue of acceptable microbial contamination levels or zero tolerance, validity of product test results relative to the methodologies used, interpretations, third party validations, what is the ALOP, the microbiological criterion including the wider implications from the recent Cadbury dilemma may well justify a review of our current sampling plan or if not the validity of current practices relative to emerging food safety demands or concerns.

Base on the comments provided by Charles C. and Wayne, this is indeed a huge subject and I believe both Charles C and Wayne have a whole lot more information and experience to share with us. Ready to receive

[Charles.C]

Dear Wayne / CharlesChew,
Such is life in the fast lane. Note that the '5% defects' value is a maximum at the selected probability, hopefully the actual situation will be lower incidence, ie nil.
These values illustrate the fact that no feasible sampling plan can ensure complete absence of a particular organism as already noted by the experts in the chocolate saga. The expert's related comment was 'End-product testing is not a suitable instrument for guaranteeing the safety of the food and a robust HACCP (Hazard Analysis Critical Control Point) needs to be in place. Unquestionably so however I wonder regarding the microbiological validation for any critical limits of such a plan.
I recommend a look at the ICMSF book Micro-organisms in Foods No2 (1986) which will do a much better job at explaining the probabilities than me.
Regarding the MPN method, you are correct in that the MPN is not a very precise estimate of the actual number of organisms present but there are few options for estimating the low levels of bacteria involved. One should exercise an appropriate degree of flexibility in assessing such quantitative micro.data which is where one difficulty in formulating microbiological criteria comes in.
Your question regarding false positives etc comes down to the particular method/application. Standard methods are validated in this respect. This ref. may clarify -

http://www.aoac.org/...2Guidelines.pdf

Rgds / Charles.C

[wayne]

Dear Wayne / CharlesChew,
Such is life in the fast lane. Note that the '5% defects' value is a maximum at the selected probability, hopefully the actual situation will be lower incidence, ie nil.
These values illustrate the fact that no feasible sampling plan can ensure complete absence of a particular organism as already noted by the experts in the chocolate saga. The expert's related comment was 'End-product testing is not a suitable instrument for guaranteeing the safety of the food and a robust HACCP (Hazard Analysis Critical Control Point) needs to be in place. Unquestionably so however I wonder regarding the microbiological validation for any critical limits of such a plan.
I recommend a look at the ICMSF book Micro-organisms in Foods No2 (1986) which will do a much better job at explaining the probabilities than me.
Regarding the MPN method, you are correct in that the MPN is not a very precise estimate of the actual number of organisms present but there are few options for estimating the low levels of bacteria involved. One should exercise an appropriate degree of flexibility in assessing such quantitative micro.data which is where one difficulty in formulating microbiological criteria comes in.
Your question regarding false positives etc comes down to the particular method/application. Standard methods are validated in this respect. This ref. may clarify -

http://www.aoac.org/...2Guidelines.pdf

Rgds / Charles.C

Dear Charles C

The ICMSF has recently (2002) published its latest version (i.e. Microorganisms in Foods No.7). You are right- all details are spelt out in the book.

The expert states that 'The stringency of a food safety system is a relative attribute'. It goes further that 'whether for attribute (i.e. presence/absence) data or microbial numbers, the setting of microbiological limits will ultimately be a matter of judgement, experience, and considering the risk and consequences of failure'. Mate (my engineering friend) - I am not sure if you do understand this statement.

No one can deny that the probability of acceptance (Pa ) is a function of n and c. Is c=0 is valid scientifically? But, mathematically or statistically zero is not in existence. To a layman, there is no different between a 0.0012 % defect (numerical number) and a 0% defect (non-numerical number).

Charles C - is zero not a numerical number within the context of risk/safety? I need help.

Regards
Wayne

[Charles.C]

Dear Wayne,
I'm not sure I fully understand yr question so I apologise if this is not a meaningful response and please revert again.
The rule of 'zero tolerance' derived from the intention that ready-to-eat foods should be totally free of salmonella (= zero percent salmonella contamination). The causes / consequences of this not being the case will vary depending on species, product, process, consumer etc but were considered sufficiently unacceptable by the 'experts' to justify the general rule. I'm not sure laymen would cheerfully eat food from a lot known to have 0.001 % salmonella contamination, I prefer to believe that consumers expect such food to have been rejected. Having said that, one comes back to the sampling problem and HACCP.
Rgds / Charles.C

[Charles.C]

Dear All,
I thought I should add a bit more detail to the above discussion since this was supposed to be a general thread -

Zero tolerance - this is a very controversial subject (I must admit I hadn't realised how widespread the topic had become for which thanks Wayne), for example see -

http://darwin.nap.ed...8X/html/24.html
above link not working, try -
http://www.nap.edu/o...d=10690&page=24
(added 19-01-2010)
and
http://ift.confex.co...paper_21732.htm

(maybe the 2nd ref was the thrust of your 'zero' query, Wayne ? You are lucky to have a copy of Mics. in Foods 7)

The ZT suitability in the case of L.monocytogenes has been sufficiently questioned so as to cause some re-thinks, for example see -

"5.2 QUESTION 1
Estimate the risk from L. monocytogenes in food when the number of organisms range from
absence in 25 grams to 1000 colony forming units per gram, or millilitre or does not exceed
specified levels at the point of consumption."
( http://www.who.int/f.../mra5_part5.pdf )

And
http://www.foodscien.../fshbull15a.htm

Moving on, if you are interested (brave enough) to see how the statistics / mechanisms of some of the top bad bugs compare, can try -

http://members.ift.o.../0/bacteria.pdf.

And a survey of control techniques are here -

http://www.ift.org/p...food_safety.pdf
note - above link not working, can try -
 managing_food_safety.pdf   205.62KB   59 downloads
(added 19-01-2010)

Finally, for a powerful viewpoint on a range of microbiological criteria and related -

http://www.fao.org/d...3e/y4743e05.htm
and
http://www.fao.org/d...3e/y4743e0n.htm

Hope this is not excessive referencing,
Rgds / Charles.C

[Charles Chew]

Dear Wayne / Charles C.

Someone appreciatively wrote a statement somewhere which I sadly could not remember but it was something like "You can have safe foods but you can never have risk-free foods"

Can this topic be summed up in this single line of statement?

I shall leave it to the expertise of Wayne and Charles C. to determine that but clearly, this topic deserves more attention. Where are all the microbiologists when you are needed?

Charles

[wayne]

Dear Charles C & Charles Chew

I was out in the bush for a week and here I am.
The info given in the reference is indeed useful and very powerful. Thank you Charles C.

Whether 'safe food = or ≠ risk-free food' is not sure but is an interesting issue.
However, I wonder whether the hypothetical results (i.e.the 2 class plan with c=0) of a process capability study presented in some of these published articles are valid or not. The argument includes:

(1) The existence of statistical error / mathematical errors /Sampling errors, disregarding sample size should not be in doubt; but it is not sure whether ZT (c=0) can exist under this environment?

(2) A bacterium is multiplying at a rate of 15-20 minutes per cycle and it appears that this dynamic factor has been ignored in the formulation of a given MC plan (correct me if I am wrong). It means that when c=0 at time x, the counts may not reflect the actual population when the samples are taken and analyzed at time x + 1 (i.e. 0.3cfu/ml at time x may have increased to a magic number at time x+1). Clearly, this factor has further complicated the issue of acceptable/non-acceptable criterion. Charles C, would you be able to re- confirm that the status of this dynamic factor has actually been considered in the plan (i.e. c=0)? I need your input on this topic or any idea that can enhance the understanding of this topic.

Risk assessment relies on the availability and validity of data. If the analytical method(s) is invalid, should one proceed with the validation & verification of the FSMS in question? What action should an auditor take?

My mate / engineering food auditor argues that ' a food safety expert is not required when audit a food safety system; it is a simple job as long as we can carry a checklist along with us and skillfully master the QMS audit technique'. Wow! It looks simple but I am not so sure mate! Can someone give his/her view on this statement?

Kind Regards
Wayne

[Charles.C]

Dear Wayne,

Yes, this is a complicated interplay of safety objectives / knowledge / prediction. Including a lot of assumptions as you say. Just like HACCP in fact. I guess the proof of the pudding has to be in .... (and the validation)
Generally foods are created / maintained in an appropriate way to prevent bacterial multiplication, eg refrigeration etc.
I think BRC requires laboratories doing analyses for pathogens to show ISO17025 capability. I would guess yr mate (I assume a non-microbiologist) considers his checklist adequate to routinely audit this as well and perhaps a good one will though I personally doubt it. Regardless, a standard list of microbiological requirements should have served him well in the recent chocolate disaster.

Rgds / Charles.C

[Charles Chew]

Risk assessment relies on the availability and validity of data. If the analytical method(s) is invalid, should one proceed with the validation & verification of the FSMS in question? What action should an auditor take?

Dear Wayne,
You have certainly raised a very realistic issue. Evaluation of Perfomance Objectives depends very much on the validity of the data implementation for Performance Analysis.

Base on your argument, the Performance Criterion seems uncertain with regards to the setting of an acceptable microbiogical sampling plan when "c" is "0" as you have rightly argued so that the proliferation rate of a bacterium is never static unless the test environment is controlled which in this case is almost always dynamic unless based on lots of assumptions as Charles C. has pointed out. So indeed, can "c" be "0".

Again, I like to stress Charles C.'s reminder of ISO 17025 test methodology as an important benchmark for .........

By the way, I think your Engineering Food Audito Mate should seriously reconsider about food auditing.

Regards
Charles Chew

[wayne]

Dear Wayne,
You have certainly raised a very realistic issue. Evaluation of Perfomance Objectives depends very much on the validity of the data implementation for Performance Analysis.

Base on your argument, the Performance Criterion seems uncertain with regards to the setting of an acceptable microbiogical sampling plan when "c" is "0" as you have rightly argued so that the proliferation rate of a bacterium is never static unless the test environment is controlled which in this case is almost always dynamic unless based on lots of assumptions as Charles C. has pointed out. So indeed, can "c" be "0".

Again, I like to stress Charles C.'s reminder of ISO 17025 test methodology as an important benchmark for .........

By the way, I think your Engineering Food Audito Mate should seriously reconsider about food auditing.

Regards
Charles Chew

Dear Charles C & Charles Chew

After reading this thread in the forum, my engineering food auditor friend has make a firm decision not go ahead with the audit of ISO22000 system even though his CB has registered him to be the lead food auditor.

He admitted that microbiological risk assessment is not as simple as what he has originally thought of; it is a complicated system. He was surprised that the authority has given him the ticket to audit ISO22000 even though he was (is) technically not qualified.

However, I encourage him not to give up but to take up a course in food microbiology or a similar field. My engineering mate wants to thank both Charles for awakening him or else he would have committed a professional crime.

Charles C, I do share with your feeling as being an auditee. Indeed, I was being audited by an "extra large international company" two days ago. I do understand your feeling as an auditee? Some auditor(s) are behaving as they were "above the law of the land" even though it doesn't make any sense out of it (i.e. with little scientific ground).

It is sad to see that some laboratories with ISO17025 capability are NOT better than those without ISO17025; their work/ reports are crap. Charles C, do/did you have similar experience in this? This again, ..... ( perhaps it is the quality work of an auditor, not my mate this time).

I have to rush off to the bush again and will be back in a couple of days. I will talk to you both soon. Take care.

Simon, would it be nice if more technical members can / are willing to make contribution to this topic? We need this to safe more life in our community.

Kind Regards
Wayne

[Simon]

Simon, would it be nice if more technical members can / are willing to make contribution to this topic? We need this to safe more life in our community.

Yes it would Wayne, I'm doing my best, but you can only lead a horse to water, you cannot make it drink.

I won't give in and please don't you - I'm sure things are destined to get better real soon.

I have some marketing plans in the pipeline.

Regards,
Simon

[just me]

Simon, would it be nice if more technical members can / are willing to make contribution to this topic? We need this to safe more life in our community.

Dear Wayne,

Hope you don't mind me chipping in, but there not too many ppl like Charles C., Charles Chew and your good self around.

Most of us are, as Simon put it "read it with awe" . Still trying to decipher the terms in setting up microbiological criteria. Heavy hands that is trying to dig back into text books and references.

Cheers,

[Simon]

Hope you don't mind me chipping in, but there not too many ppl like Charles C., Charles Chew and your good self around. Most of us are, as Simon put it "read it with awe" . Still trying to decipher the terms in setting up microbiological criteria. Heavy hands that is trying to dig back into text books and references.

Aghh! Please don't offend the experts JM.

Rather than too many I would be happy to see many more experts. We can learn from them and we need them to answer the questions posted by numpties like me.

Sorry to derail your thread guys please continue.

Simon

[just me]

Aghh! Please don't offend the experts JM.

Apologies.

[Charles.C]

Dear All,
I noted that Simon in another thread mentioned he had followed this one with interest but some bewilderment.
I am not at all surprised. This subject has developed chronologically via some very tortuous paths to the present and the story is undoubtedly not over yet. I thought a brief summary might be stimulating however it became slightly lengthy so I uploaded it.

Rgds / Charles.C
 MC_comments.doc   27.5KB   56 downloads

note - above original attachment corrupted

try -  MC_comments.doc   27.5KB   19 downloads
(added 19-01-2010)

[Charles Chew]

I thought it would be interesting to look at some of the extracts from the sub-title of “The Concept of Zero Tolerance” – Microorganism in Foods 7:

This could well throw the usual sampling plan model out of the windows. At least four considerations challenge the ideal of zero tolerance.

a. No feasible sampling plan can ensure a complete absence of pathogen. Even when c = 0, it cannot be guaranteed that the lot is completely free of the organism no matter how large the sampling units n. However, what can be stated is the probability of acceptance (Pa) for lots of various qualities as a function of n and c………………for example if n = 60 and c = 0 with (Pa) = 0.5 meaning there is one chance in two that a lot will be accepted when 1% of the sample units are defective.

b. Plans in which c = 0 are not necessarily the most exacting. For example, if one sets a limit of 5% defectives in a lot, the plan n = 95, c = 1 will accept unacceptable lots less often, and acceptable lot more often than will the plan be n = 60 and c = 0. In other words, the former plan is more discriminating, even though it admits the presence of pathogen. Preference is for c = 0 is influenced by the wish to emphasize that absence id the desired objective (unlikely to be guaranteed) and by the knowledge that pathogens such as Salmonella may be found by direct examination of certain foods and the findings cannot be ignored.

c. The sampling plan typically used for lot acceptance testing assume that defects (eg Salmonella) are randomly distributed through the lot. This may be difficult or impossible to achieve.

d. It is not yet commercially possible to market foods completely free from pathogen. A sampling plan adjusted to this situation would seem more realistic and satisfactory than one based on an unattainable ideal of complete absence (i.e. zero tolerance)

After all this, we may have to remove c = 0 from all our sampling plan

Edited by charleschew, 29 August 2006 - 05:04 PM.

[Charles.C]

Dear CharlesChew,

Yr efforts to prolong this thread should be acknowledged.
I would guess that the statistical-related items which you have extracted are probably accepted by most interested parties as such. However they do not take account of various ‘psychological' factors which seem to dominate unless clear evidence to support a change from the use of a zero tolerance requirement exists such as in the recent case of L.monocytogenes in ready-to-eat items.

To illustrate such factors I have extracted two pieces from FS papers at a recent symposium (2005) (Dr. C.de W. Blackburn and Prof.P.G.Wall respectively) -

(pg 106)
'While on this subject, it is important to realise that risk assessment should not be considered the sole preserve of the scientist. However, risk perception research indicates that lay people perceive risks differently from experts. Scientists focus on probability and severity of harm while the lay public respond additionally to other (outrage) factors associated with the situation: the distribution of benefits and risks; fairness of the risk; degree of personal control over the exposure; voluntary or involuntary exposure; severity of consequences; and characteristics of the person at risk (children generating greatest feelings of protection). For example, whereas governments and industry generally support the adoption of new technologies, consumer organisations question the underlying need, which results in less willingness to accept the risks, even when they are small (Macfarlane, 2002). Overall, it would seem that the scientific evaluation that a risk is small is not the principle determinant of the consumer position. If they are not convinced of the utility or need, they are not willing to accept the risk, however small, especially when they consider that benefits accrue mostly to industry (Macfarlane, 2002).'

And (Pg118)

'Although the food supply has probably never been safer and there are more regulations than ever before and most countries have food safety agencies to co‐ordinate controls, public fears appear to have increased substantially. Some risk sources can create social and economic impacts that are far greater than would be predicted on the basis of estimated direct harm. These hazards, for example radiation, or recombinant DNA research, generally evoke a concern and dread among segments of the public in some countries, amplifying the adverse impacts of these technologies and evoking a risk management response that is often in proportion to the media interest rather than in proportion to the degree of risk to the consumer. A lay person's concerns about risk are sensitive to such characteristics as the voluntariness of exposure, the potential for catastrophe, the newness of the technology, uncertainty of outcomes, impacts that are perceived to be inequitably distributed across groups and when the management of the issue is brought into question.'

The latter article also contains some critical comments regarding the application and value of legislative directives vis-a-vis the food industry - makes interesting reading IMO, along with all the other items in this volume.

The link is -
http://www.eu-rain.c...M6ProceedBW.pdf

Rgds / Charles.C

[Charles.C]

Dear All,

A number of queries have occasionally been posted regarding the possible effect of measurement error in microbiological results particularly as compared to regulatory limits.

This link (eg see pages 10, 15) nicely illustrates the potential for rather large surprises in this respect (surprised me anyway) -

http://www.foodstand...c20039annex.pdf.
above link not working, see attachment below -

 UNCERTAINTY_SANCO_wpcc20039annex.pdf   109.03KB   25 downloads
(added 19-01-2010)

The significance of such variations appears to be currently under discussion, perhaps already concluded although was unable to find later references than this -

http://www.flep.org/...entversion3.doc.



Rgds / Charles.C

[wayne]

Dear Charles C. & Charles Chew


Thanks to both of you by keeping this thread going. Whether we like it or not, it is one of the most important elements in risk assessment - should we say in the ISO22K implementation. My engineering friend has totally lost in this aspect..

I am not lost. Just to let you both know that I was away to the bush and am still in the bush. I will response to both of your comments in a couple of days after returning to civilisation.

Best Wishes & talk to you both soon.
Wayne

[Charles.C]

Dear Wayne,

Nice to hear from you. Watch out for the spears and blowpipes.

You may have noticed that the official discussion of errors linked in the previous post seemed to not include much on the subject of detection limits for pathogens.
Since the typical regulatory requirements for ,say salmonella, are 'not to be detected in a 25g sample(s)' the techniques are required to be capable of (correctly) detecting 1 salmonella in 25g. There are many validated procedures / technologies for Salmonella detection and without going into details, the basic answer for these seems to be yes, but with some reservation (on my part) regarding unfavourable factors like food matrix, level / type of background flora.
As an example only, can see -

http://www.biogenome...Cert Report.pdf.

Rgds / Charles.C

[wayne]

Dear Charles C.

It is nice to talk to you again. I must apologise for not being able to reply you sooner.

PCR is a good & rapid method for microbe confirmation or validation of the validity of the selected critical limit(s) for the identified CCP(s). However, it has its weakness/uncertainty in term of its sensitivity & accuracy. It picks up dead cells as well; this is not what we want for validation. What we need is the culturable living cells. Again, will the chocolate Salmonella syndrome be repeated if we accept a 1% error (false (+) /false (-) in the PCR micro-analysis?

Secondly, we are not sure whether the info given in FDA method was derived from laboratory samples ( experimental samples) or environmental samples. Furthermore, performance analysis for laboratory variations (inter & intra laboratory comparison) has appeared not been conducted (correct me if I read it wrongly).

We all know that PCR recovery ratio differs according to food matrix. Also the presence of PCR inhibitor(s) or non-target DNA in an environmental sample can further complicate PCR signals.


Regards/Wayne