16 replies to this topic

[earthbornstew]

Hi Everyone,

I hope someone can help me with this. I work with a trader in milk powder and help with developing specifications. We have two tests which state a max limit of <1 CFU/g for e.coli ( test method iso 11866-1) and coliforms (iso 4831/4832). Previously these specs stated <1 CFU/0.1g max.

However I don’t believe this is correct, as the testing will usually be carried out by adding 1g milk powder with 9mls diluent to reconstitute the samples, and plated with 1ml. Would I be correct in saying that the lowest reportable result would actually be <10 CFU/g and that reporting <1 for zero detected growth is incorrect?

If not, how would you be able to test for that in milk powder?

My background isn’t very thorough in micro. I’ve tried reading ISO7218 requirements and guidance for micro exams but it doesn’t explain to me clearly my question above.

Thanks for any help on this.


Sent from my iPad using Tapatalk

[earthbornstew]

Additional question on above, is <10 CFU/g equivalent to <1/0.1g?


Sent from my iPad using Tapatalk

[bensmith007]

Hi,

 

This is just a simple dilution factor calculation. Often in micro testing you need to use dilutions in order to get a workable number of bacteria growing on the petri dish. If you just try and plate the undiluted plate for something with a lot of bacteria (e.g. a root vegetable) then there are too many bacteria colonies to be able to easily count. In this case you plate a series of dilutions (generally 1 in 10 as you describe above- 1g milk powder plus 9g (or ml) of water is 1 in 10) of decreasing strength until you obtain a plate with a countable number of bacterial colonies on it- between 3 and 50 colonies normally (don't quote me exactly on those numbers). To relate back to the number of bacteria that this represents per gram of milk powder, you need to multiply the number of bacteria by the dilution factor.

 

In the case you describe above, the microbiologist would take the number of colonies grown on the dish and multiply that by 10 to give you CFU/g and thus this will account for the 1 in 10 dilution of the milk powder initially performed.

 

That result will usually be in the units of CFU/g as industry standard, and the actual value will depend on how many bacterial colonies grew on the plate. In order to reach your limit of <1 CFU/g, no bacterial colonies should be found growing on the plate. If one colony was found, then your actual result would be 10 CFU/g after multiplying by the dilution factor.

 

I suggest you speak to the lab conducting the micro testing for you and ask them if they can give you a run down of the process or better yet go visit them and see it done for real. It's not always intuitive.

 

Suffice to say, your COA limit and reporting sounds fine, although I have no idea what lab you are using so I can't vouch for them.

 

I hope that makes a little more sense!

 

Ben

[earthbornstew]

Hi Ben

Thanks for the answer above. I do have one question on one piece you mentioned.

That result will usually be in the units of CFU/g as industry standard, and the actual value will depend on how many bacterial colonies grew on the plate. In order to reach your limit of <1 CFU/g, no bacterial colonies should be found growing on the plate. If one colony was found, then your actual result would be 10 CFU/g after multiplying by the dilution factor.

I would wonder if you could say <1cfu/g though if no growth was detected but you were using different dilutions, say from one batch to another.

Below is an image from a piece I found online about interpretation of results that would say you should include the dilution factor, even if no detection. So the lowest reportable could be <10 for a 1:10 dilution, <100 for 1:100 ...... etc. it’s from http://www.microchem...ion-of-results/




Sent from my iPad using Tapatalk

[Charles.C]

Hi Earthbornstew/Ben,

 

ISO are very thorough but always make things so complicated to follow.

 

The simple case is basically arithmetic until you get involved with averages when you hv to examine all the details.

 

Example 1g sample product mixed with 9 ml liquid.This is equivalent to 1g product in 10g solution.

This implies 1ml solution ~ contains 0.1g product.

 

Assume 1ml of solution is transferred to agar plate and after incubation can detect 1 colony.

This implies there is 1cfu in 0.1g product which is equivalent to saying there is 10 cfu in 1g of product, ie 10cfu/g

 

Therefore, zero detection of colonies implies < 10cfu/g product.

Edited by Charles.C, 11 December 2017 - 11:32 PM.
re "formatted" slightly

[Charles.C]

addendum

 

to be able to state that <1 cfu/g for a solid, probably need to use a MPN technique involving multiple tubes.

 

I assume the procedure you refer does not include a MPN variation.

[bensmith007]

Interesting- I've not come across that before in my education or my (relatively brief) time in industry.

 

Check out slide 26 of this Powerpoint from FSIS: https://www.fsis.usd...pdf?MOD=AJPERES

which makes the point that the portion of test sample will dictate the theoretical limits of detection.

 

Does this explain why a lot of specifications will not say 'less than X per gram' or whatever, but rather 'ABSENCE of pathogen in X grams of food sample'?

 

I am at the limit of my knowledge here, so I suggest finding someone more microbiology oriented on this forum, or just speak to your pathogen testing lab so they can explain!

 

Ben

[earthbornstew]

Hi Charlie

Thank you for the input above.

What I’m taking from everything so far is that whether we state <10cfu/1g or <1cfu/0.1g , these should be equivalent. It seems that only when the spec states <1cfu/g that they are actually incorrect.

Is there any requirement (Codex, ISO...etc) which would indicate whether <10cfu/1g or <1cfu/0.1g would be the preferred reporting format?


Sent from my iPad using Tapatalk

[Charles.C]

Interesting- I've not come across that before in my education or my (relatively brief) time in industry.

 

Check out slide 26 of this Powerpoint from FSIS: https://www.fsis.usd...pdf?MOD=AJPERES

which makes the point that the portion of test sample will dictate the theoretical limits of detection.

 

Does this explain why a lot of specifications will not say 'less than X per gram' or whatever, but rather 'ABSENCE of pathogen in X grams of food sample'?

 

I am at the limit of my knowledge here, so I suggest finding someone more microbiology oriented on this forum, or just speak to your pathogen testing lab so they can explain!

 

Ben

 

Hi Ben,

 

Indeed, the sample size is directly related to detection sensitivity, eg some texts use 25g, others are more strict and require 50g which will increase chance of detection.

 

The terminologies used relate as to whether the testing procedure is qualitative or quantitative.

 

ABSENCE in 25g sample means that using a (hopefully) standard qualitative procedure, the pathogen was not detected in 25g. (Absence per se in a product lot is impossible to prove due statistical/analytical sampling errors).

 

As in yr link MPN techniques are quantitative so allow a numeric expression, eg <3 cfu/g where the detection limit can be lower than direct plating techniques but also likely require a lot more work.

 

Since most/all specs have a zero tolerance for pathogens like salmonella the labs normally only use a qualitative procedure for this.

In comparison S.aureus usually has a finite limit in specs so a quantitative procedure is employed, either direct plating or MPN.

[Charles.C]

Hi Charlie

Thank you for the input above.

What I’m taking from everything so far is that whether we state <10cfu/1g or <1cfu/0.1g , these should be equivalent. It seems that only when the spec states <1cfu/g that they are actually incorrect.

Is there any requirement (Codex, ISO...etc) which would indicate whether <10cfu/1g or <1cfu/0.1g would be the preferred reporting format?


Sent from my iPad using Tapatalk

 

Hi earthbornstew,

 

It is more usual to use 1g as a reference. eg <10cfu/g. This format is typically used in Product Specifications.

 

<1cfu/g is a possible Product Specification (obviously more strict than <10cfu/g) but will likely require a MPN technique if product is a solid. (Or perhaps via use of mutiple Plates if direct plating).

 

For a liquid, testing compliance to such a spec. is directly achievable using a 1ml sample (zero diluent) and the Maths in my Post 5 (However sometimes the liquid is too viscous etc so that a diluent must be used)

[earthbornstew]

Hi Everyone

Just a quick update. We’ve reviewed this and discussed with some of our suppliers and they’ve agreed going with <1 CFU/0.1g.

Thanks for all the advice.


Sent from my iPad using Tapatalk

[KHC]

Hi Earthbornstew/Ben,

 

ISO are very thorough but always make things so complicated to follow.

 

The simple case is basically arithmetic until you get involved with averages when you hv to examine all the details.

 

Example 1g sample product mixed with 9 ml liquid.This is equivalent to 1g product in 10g solution.

This implies 1ml solution ~ contains 0.1g product.

 

Assume 1ml of solution is transferred to agar plate and after incubation can detect 1 colony.

This implies there is 1cfu in 0.1g product which is equivalent to saying there is 10 cfu in 1g of product, ie 10cfu/g

 

Therefore, zero detection of colonies implies < 10cfu/g product.

Hi Charles, I would like to ask your opinion regarding this.

Example 10g sample product mixed with 90 ml liquid then transfer 1ml for plating. If no growth on the plate, we will report <10cfu/g right? How about if transfer 10ml for membrane filtration and there is no growth on membrane. Then how do i report the result? still <10cfu/g? <10cfu/10g? <1cfu/g? I suddenly stuck with this question that it never come out in my mind until today.

[Charles.C]

Hi Charles, I would like to ask your opinion regarding this.

Example 10g sample product mixed with 90 ml liquid then transfer 1ml for plating. If no growth on the plate, we will report <10cfu/g right? How about if transfer 10ml for membrane filtration and there is no growth on membrane. Then how do i report the result? still <10cfu/g? <10cfu/10g? <1cfu/g? I suddenly stuck with this question that it never come out in my mind until today.

Hi KHC,

 

I think you have a "case" of Apples and Oranges.

 

Can study this attachment for membrane filtration calculations -

 

 MF Counting Rules.pdf   40.87KB   18 downloads

(eg example 12)

 

A lot of hypotheticals but if i do the tortuous maths correctly, referring to the original sample gives <1cfu/gram, ie the MF wins on sensitivity,

 

In reality, first dilutions are often too viscous to be plated and i anticipate that membrane filters also have physical limitations.

[KHC]

Hi KHC,

 

I think you have a "case" of Apples and Oranges.

 

Can study this attachment for membrane filtration calculations -

 

MF Counting Rules.pdf

(eg example 12)

 

A lot of hypotheticals but if i do the tortuous maths correctly, referring to the original sample gives <1cfu/gram, ie the MF wins on sensitivity,

 

In reality, first dilutions are often too viscous to be plated and i anticipate that membrane filters also have physical limitations.

Hi Charles, actually this came out after i read the usp method for nonsterile products. I don't quite understand for the membrane filtration part.

eg. sample preparation stated use 10g or 10ml sample and dissolve or dilute in 1:10 dilution. For membrane filtration, it stated transfer suitable quantity of the sample (preferably representing 1g of product or less if high cfu expected) to the membrane filter. So in this case, 1ml of the 100ml diluted sample (10g sample with 90ml diluent) will do?

Besides, the method also mentioned when examining transdermal patches, separately filter 10% of the sample volume. So, i need to filter 10 times of 1ml each and get the average reading to represent cfu/g and also fulfill the 10% sample volume if the total volume of prepared sample is 100ml?or I straight filter 10ml? if 10ml then i will get confused how to report the result if no growth on the filter since the sample is the diluted in 1:10 at first place.
 

[Charles.C]

Hi Charles, actually this came out after i read the usp method for nonsterile products. I don't quite understand for the membrane filtration part.

eg. sample preparation stated use 10g or 10ml sample and dissolve or dilute in 1:10 dilution. For membrane filtration, it stated transfer suitable quantity of the sample (preferably representing 1g of product or less if high cfu expected) to the membrane filter. So in this case, 1ml of the 100ml diluted sample (10g sample with 90ml diluent) will do?

Besides, the method also mentioned when examining transdermal patches, separately filter 10% of the sample volume. So, i need to filter 10 times of 1ml each and get the average reading to represent cfu/g and also fulfill the 10% sample volume if the total volume of prepared sample is 100ml?or I straight filter 10ml? if 10ml then i will get confused how to report the result if no growth on the filter since the sample is the diluted in 1:10 at first place.
 

Hi KHC,

 

Perhaps you could post the referred USP method ? (Do they not offer an appropriate formula/explanation?)

Regret no familiarity with transdermal patches.

[KHC]

Hi KHC,

 

Perhaps you could post the referred USP method ? (Do they not offer an appropriate formula/explanation?)

Regret no familiarity with transdermal patches.

Hi Charles,

Here is the link

https://www.usp.org/...a_34_6_2008.pdf

[Charles.C]

Hi Charles,

Here is the link

https://www.usp.org/...a_34_6_2008.pdf

Hi KHC,

 

Thks link.

 

Regarding yr 1st query, the criterion is to enable approx. the stated target number of growth colonies so apparently answer = yes (have no direct experience with membrane filters).

 

Sorry but I was unable to fully understand the text associated with transdermal patches. You probably need to access a Pharmaceutical Micro textbook for a more detailed explanation of the Procedure.