Jump to content

  • Quick Navigation
Photo

Result for Total Plate Count of different sample/diluent ratio used

Share this

  • You cannot start a new topic
  • Please log in to reply
8 replies to this topic
- - - - -

Asiah

    Grade - Active

  • IFSQN Active
  • 4 posts
  • 1 thanks
0
Neutral

  • Malaysia
    Malaysia

Posted 05 March 2013 - 08:45 AM

Hi all, sorry for my limited English. i would like to have an answer for this question. Can we get the same result if we used different total of sample/diluent, as below:


a. 25 g sample/225 ml diluent (10-1) ------- 10 ml sample/90 ml diluent (10-2) ------- 10 ml samle/90 ml diluent (10-3)


b. 5 g sample/45 ml diluent (10-1) ------- 1 ml sample/9 ml diluent (10-2) ------- 1 ml samle/9 ml diluent (10-3)


c. 10 g sample/90 ml diluent (10-1) ------- 10 ml sample/90 ml diluent (10-2) ------- 10 ml samle/90 ml diluent (10-3)


if it will give us different CFU, please give a clear explanation.

Thank you very much.

Asiah


Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 20,542 posts
  • 5662 thanks
1,544
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 06 March 2013 - 07:31 AM

Dear Asiah,

Method (a) is I think a typical textbook procedure. In principle yr dilution methods (b,c) look acceptable to me but I noticed this comment –

The size of sample to be taken will depend upon the amount of primary sample available and its homogeneity, or otherwise. Ideally, the analytical sample should never be less than 10g and it should be weighed to the nearest 0.1g into a sterile container on a top-pan or other suitable balance.


Can we get the same result ?
If it will give us different CFU?


If you mean - will all plates have identical counts ?, the almost certain answer is No unless plates have no growth or are covered with colonies.

Replicate analyses on a single sample, or analyses of replicate samples, will always show a variation among results. The reason is the unavoidable existence of “error” due to sampling or measurement.

% Total error = +/- SQRT (A^2 + B^2 + C^2)
where A = % Sampling error; B = % Distribution error; and C = % Dilution error.

If one can assume for normal routine purposes that the following percentage errors occur, can derive estimates for the typical overall error of the colony count method. Assuming a sampling error of +/- 5% (A), a distribution error of +/- 10% (100 colonies counted) (B), and a dilution error of +/- 5.5% © then the overall % error is given by +/- 12.46%
So if it is assumed that the distribution of organisms follows a Poisson series (i.e. that a hypothesis of randomness is not rejected), then the approximate 95% confidence limits on the count of 100 x 10^6 would be +/- 25%.

Some typical results are given here –

Attached File  plate count confidence limits.png   75.57KB   29 downloads

The confidence limits for dilution schemes (b,c) will increase compared to (a), I think.

Rgds / Charles.C

Kind Regards,

 

Charles.C


Asiah

    Grade - Active

  • IFSQN Active
  • 4 posts
  • 1 thanks
0
Neutral

  • Malaysia
    Malaysia

Posted 07 March 2013 - 12:42 AM

Dear Asiah,

Method (a) is I think a typical textbook procedure. In principle yr dilution methods (b,c) look acceptable to me but I noticed this comment –





If you mean - will all plates have identical counts ?, the almost certain answer is No unless plates have no growth or are covered with colonies.

Replicate analyses on a single sample, or analyses of replicate samples, will always show a variation among results. The reason is the unavoidable existence of “error” due to sampling or measurement.

% Total error = +/- SQRT (A^2 + B^2 + C^2)
where A = % Sampling error; B = % Distribution error; and C = % Dilution error.

If one can assume for normal routine purposes that the following percentage errors occur, can derive estimates for the typical overall error of the colony count method. Assuming a sampling error of +/- 5% (A), a distribution error of +/- 10% (100 colonies counted) (B), and a dilution error of +/- 5.5% © then the overall % error is given by +/- 12.46%
So if it is assumed that the distribution of organisms follows a Poisson series (i.e. that a hypothesis of randomness is not rejected), then the approximate 95% confidence limits on the count of 100 x 10^6 would be +/- 25%.

Some typical results are given here –

Attached File  plate count confidence limits.png   75.57KB   29 downloads

The confidence limits for dilution schemes (b,c) will increase compared to (a), I think.

Rgds / Charles.C


thanx for your feedback.
actually my problem is i got different result when i used (a ) method, (10-4), but when other personnel did the test used (b) method, she only got 10-2.

i wonder if we used less amount of sample we will get lower count.




Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 20,542 posts
  • 5662 thanks
1,544
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 07 March 2013 - 02:33 AM

thanx for your feedback.
actually my problem is i got different result when i used (a ) method, (10-4), but when other personnel did the test used (b) method, she only got 10-2.

i wonder if we used less amount of sample we will get lower count.


Dear Asiah,

Sorry but I didn’t understand all yr post.
I am not clear how much difference in results between yourself and your colleague.

What kind of food ? Solid / Liquid ? Frozen samples ?
Did your sample and your colleague’s sample come from same production day / batch ?
What were final results for you and your colleague ?
Can you tell me how many colonies counted on plate(s) and the dilution(s) for both people ?

Rgds / Charles.C

Kind Regards,

 

Charles.C


Asiah

    Grade - Active

  • IFSQN Active
  • 4 posts
  • 1 thanks
0
Neutral

  • Malaysia
    Malaysia

Posted 07 March 2013 - 04:21 AM

Dear Asiah,

Sorry but I didn’t understand all yr post.
I am not clear how much difference in results between yourself and your colleague.

What kind of food ? Solid / Liquid ? Frozen samples ? Solid sample (coconut square)
Did your sample and your colleague’s sample come from same production day / batch ? yes both sample from same production date and batch
What were final results for you and your colleague ? my result (method a) - 3.0 x 10-4), her result (method b) - 5.0 x 10-2)


Can you tell me how many colonies counted on plate(s) and the dilution(s) for both people ?



10 -1

10 -2

10 -3

Method a

tntc

tntc

30

Method b

50

4

ND




Rgds / Charles.C

Dear Charles, please see the red reply. thank you very much.

Regards
Asiah


Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 20,542 posts
  • 5662 thanks
1,544
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 07 March 2013 - 05:52 AM

Dear Charles, please see the red reply. thank you very much.

Regards
Asiah

Dear Asiah,

Thks data. Some first thoughts –

(1) I am not familiar with this product but IMEX with other natural items, it is normal to expect substantial variations in plate count results, eg up to 50% or more, due to natural variations within the sample and the often unavoidable limitations of the analysis procedure. The difference in yr results is however much more than usual of course.

(2) Note that sample size in method (b) may simply be too small for reliable results. I don’t know since I have always used standard method, eg BAM.

(3) Assuming the results are in units of cfu/gram and plate inoculation was 1ml, your calculations for plate count seem incorrect. Or perhaps you are not using pour plates ?
I think the results should be 30,000 cfu/gm and 500 cfu/gm respectively. Typing error ?.

(4) I guess no duplicate plates were used. This step is always highly recommended.

(5) data in method (a) looks unreliable since assuming plate of 10(-3) was read correctly, the result for plate 10(-2) should be approx. 300 count which is normally not close to tntc. This is one reason for including duplicate plates.

(6) Counts in method (b) very low for accurate result (but maybe this result well below any maximum acceptable value so you don't care ?). Even more reason to use duplicate plates if want accurate result. I hope there was no problem to read plate since IMEX, food particles can sometimes interfere at this dilution. IMO, even if procedure is proven to be working correctly, should use a larger sample to get higher counts if you are unsure of the normal bacterial level. If you want a quick, approximate method for internal use and you can validate that it is working properly, maybe useful.


Comments

(A) The result from method (b) looks very low unless there is a processing step involved to lower bacterial levels. What is previous experience this product, eg typical maximum levels of total plate count. [I wondered if your reason for using 5g sample was because previous measurements with 25g always gave results in 30,000 range up so you were trying to reduce your workload by using smaller sample? ] [some people use a reduced inoculum, eg 0.5ml, for similar purpose although this can give problems also IMEX].

(B) The result from method (a) is inconclusive due mis-match between data for dilutions 10(-2) and 10(-3). IMO duplicate plates are necessary and include dilution 10(-4) for more confidence in results while you are checking procedure.

© My suggestion is for both operators to repeat the test with duplicate plates and try a standard method like BAM using 25g sample. And include dilution 10(-4). And maybe consider sending duplicate sample to a private lab if you cannot agree. Of course, if the standard is max 100,000 cfu /gram, maybe you don’t care.

IMEX, the plate count procedure can be very sensitive to method, homogenisation method, agar media / preparation, and individual operators technique. I have previously used many non-standard methods but often had customer problems, especially when my results did not match 3rd party labs. Eventually I went back to using more standard procedures which mainly solved the problem but usually take longer time.

Hope the above is useful.
Interested to hear your feedback.

Rgds / Charles.C

Kind Regards,

 

Charles.C


Asiah

    Grade - Active

  • IFSQN Active
  • 4 posts
  • 1 thanks
0
Neutral

  • Malaysia
    Malaysia

Posted 07 March 2013 - 07:17 AM

Dear Asiah,

Thks data. Some first thoughts –

(1) I am not familiar with this product but IMEX with other natural items, it is normal to expect substantial variations in plate count results, eg up to 50% or more, due to natural variations within the sample and the often unavoidable limitations of the analysis procedure. The difference in yr results is however much more than usual of course.

(2) Note that sample size in method (b) may simply be too small for reliable results. I don’t know since I have always used standard method, eg BAM.

(3) Assuming the results are in units of cfu/gram and plate inoculation was 1ml, your calculations for plate count seem incorrect. Or perhaps you are not using pour plates ?
I think the results should be 30,000 cfu/gm and 500 cfu/gm respectively. Typing error ?. yes actually the result 30000cfu/g and 500cfu/g

(4) I guess no duplicate plates were used. This step is always highly recommended. actually the results are the average of duplicate plates

(5) data in method (a) looks unreliable since assuming plate of 10(-3) was read correctly, the result for plate 10(-2) should be approx. 300 count which is normally not close to tntc. This is one reason for including duplicate plates. i thought the plate with more than 250 count per plate should be considered as TNTC.

(6) Counts in method (b) very low for accurate result (but maybe this result well below any maximum acceptable value so you don't care ?). Even more reason to use duplicate plates if want accurate result. I hope there was no problem to read plate since IMEX, food particles can sometimes interfere at this dilution. IMO, even if procedure is proven to be working correctly, should use a larger sample to get higher counts if you are unsure of the normal bacterial level. If you want a quick, approximate method for internal use and you can validate that it is working properly, maybe useful.


Comments

(A) The result from method (b) looks very low unless there is a processing step involved to lower bacterial levels. What is previous experience this product, eg typical maximum levels of total plate count. [I wondered if your reason for using 5g sample was because previous measurements with 25g always gave results in 30,000 range up so you were trying to reduce your workload by using smaller sample? ] [some people use a reduced inoculum, eg 0.5ml, for similar purpose although this can give problems also IMEX]. i use method (a) for every samples for this product, never use method (b )and the result were varies, from no detection ( ND) at all to 30000cfu/g ++, so i think this method was more reliable. the other personnel was our new staff, and all her results were below than 1000cfu/g.

(B) The result from method (a) is inconclusive due mis-match between data for dilutions 10(-2) and 10(-3). IMO duplicate plates are necessary and include dilution 10(-4) for more confidence in results while you are checking procedure.

© My suggestion is for both operators to repeat the test with duplicate plates and try a standard method like BAM using 25g sample. And include dilution 10(-4). And maybe consider sending duplicate sample to a private lab if you cannot agree. Of course, if the standard is max 100,000 cfu /gram, maybe you don’t care.

IMEX, the plate count procedure can be very sensitive to method, homogenisation method, agar media / preparation, and individual operators technique. I have previously used many non-standard methods but often had customer problems, especially when my results did not match 3rd party labs. Eventually I went back to using more standard procedures which mainly solved the problem but usually take longer time.

Hope the above is useful.
Interested to hear your feedback.

Rgds / Charles.C


Dear Charles,


Your explanation were very useful. the real problem here, my result were out of spec, but the other result were within product spec. we already sent the sample to external lab to do the analysis. i found your explanation help me very much as i have to explain it to my superior.

Regards
asiah


Thanked by 1 Member:

Charles.C

    Grade - FIFSQN

  • IFSQN Moderator
  • 20,542 posts
  • 5662 thanks
1,544
Excellent

  • Earth
    Earth
  • Gender:Male
  • Interests:SF
    TV
    Movies

Posted 07 March 2013 - 08:56 AM

Dear Charles,


Your explanation were very useful. the real problem here, my result were out of spec, but the other result were within product spec. we already sent the sample to external lab to do the analysis. i found your explanation help me very much as i have to explain it to my superior.

Regards
asiah

Dear Asiah,

Thks for response. I can understand the reason for your interest.

Since you are concerned with compliance to a standard, it increases the reason to use a standardised method IMO.

The individual results for duplicate plates should give you some idea of the method variability.

Regarding TNTC, the interpretation has varied with time and fashion :smile: . Currently, as you say, BAM quotes 250, previously it defined it as a plate having a colony count "significantly beyond count range of 300". Basically it depends on things like how much statistical error you can accept plus allowing for possible bacterial errors due overcrowding on the plate. No method is perfect for all situations, I generally use the (expanded) plate evaluating procedure as described in Compendium of Methods for the Microbiological Examination of Foods (available on-line) which i think is more informative than BAM.

I was surprised at the range of results you are finding. ND is highly unlikely IMO.
Even more surprised that your standard’s maximum is less than 30,000 cfu/gram. Is this is some kind of cooked presentation.?

IMEX, depending on the bacterial variability of your product / the actual spec./ the external lab’s procedure and skill, and most of all the actual nature of the sample, their results may / may not agree with you. (I certainly find it hard to believe that a natural, raw, tropical product is consistently giving plate counts below 1000cfu/gm unless some bactericidal process has been applied to it. But it's far more difficult, make it impossible, to predict what level of bacteria might be present when ignorant of the process :smile: )
My usual method when sending samples to external labs is to add one appropriate dummy sample if possible which has been well analysed to act as a check on the lab, but this is not always easy to find.
You might also consider repeating the sampling / analysis if the material is still available.
Good luck.

Rgds / Charles.C

Kind Regards,

 

Charles.C


Thanked by 1 Member:

RMAV

    Grade - SIFSQN

  • IFSQN Senior
  • 407 posts
  • 122 thanks
44
Excellent

  • United States
    United States
  • Gender:Male
  • Location:USA - Midwest
  • Interests:QA, Micro, Sanitation;
    Meats, Juice, Condiments;
    SQF, Audit, and aviation

Posted 08 March 2013 - 04:53 AM

Asiah, thank you for bringing a real-world example of challenges we face in the micro lab and for following up on the discussion. This is so helpful.


Charles C., you continue to be awesome!


Asiah,

If you could post it, I would be interested in the results and conclusions you draw from the external testing.





Share this

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users