# Distribution of microorganism in food/ dairy products

Best Answer Charles.C, 26 August 2017 - 05:40 AM

Hi MDG,

(1) I presume yr stated limit of coliform <10cfu/g  refers to the finished product and therefore, in practice, an accumulated "lot" of the product. It is unclear what/where yr own sampling is being carried out.

If this limit is associated with some particular Standard, it is common to find that the Standard also defines the sampling/analytical procedure used to demonstrate compliance.

IMEX, it is more common to find micro. lot acceptance requirements stated in a nmcM format. For example I noted an ice cream standard elsewhere of the form 5/10/2/100. A sole criterion such as <10cfu/g is Ok as a guideline target but sort of statistically limited from an acceptance POV unless it refers to an average result based on a defined number of samples.

(2a) There is an alternative way to utilise data from yr OP via either basic statistics or a process POV.

If you assume that yr set of data are taken from a process "in control", the average/std.dev. of the data can be used to predict the probability that, for the subsequently accumulated batch, any random sample from the batch will yield a result  <10cfu/g (ie compliance with spec) (assuming same analytical procedure).  Conversely, you can estimate the necessary target average required if the first calculation gives too low a probability of compliance. This uses basic statistics formulae and should be quite easy to apply.

As an extension of (2a) you could consider implementing a process control procedure (again basic textbook stuff) -

(2b) Assuming that you set a target average coliform process level to be such that, theoretically, any random sample of the subsequent accumulated lot will  "likely" match yr OP-spec (ie <10cfu/g), yr timed sampling data can be used to test whether yr process is in suitable "statistical control" to consistently match such a requirement.

This means that you set yr target process average level of coliform at a certain value, say X (below 10cfu/g), such that a random sample from the resulting batch will have, for example, an approx 99% probability of giving a result of < 10cfu/g (3 sigma outer control limit).

This requires (a) defining X from an estimated Process std. deviation (via sampling data)  and then (b) using timed sampling data to test whether the process data conforms to the required  control limit. Usually this procedure requires at least 2 samples being analysed at each designated time.

The above set-up is fully described in any textbook on process statistical control. Once you've done the initial calculations, you just plot the points with time.

One caveat to the above is that inaccurate analytical data may generate large standard deviations so that the adequacy of process control becomes difficult to evaluate.

Go to the full post

12 replies to this topic

### #1 MDG

MDG

• IFSQN Member
• 98 posts
• 5 thanks
1
Neutral

• India

• Gender:Male
• Location:India

Posted 19 August 2017 - 02:07 AM

Hello,

Recently we have observed a uneven distribution of coliform in the food product, The sample was drawn at every one hour interval   But result indicate  a uneven distribution of organism . The initial count was 10 and results of every hour  was found 0 , 10, 20, 0, 40,40,0 .  The microbiologist is analysing the sample in duplicate and with blank as a control.

The question is why this type of variation and any how to interpret the result.

Thank you,

### #2 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 19 August 2017 - 03:48 AM

Hello,

Recently we have observed a uneven distribution of coliform in the food product, The sample was drawn at every one hour interval   But result indicate  a uneven distribution of organism . The initial count was 10 and results of every hour  was found 0 , 10, 20, 0, 40,40,0 .  The microbiologist is analysing the sample in duplicate and with blank as a control.

The question is why this type of variation and any how to interpret the result.

Thank you,

Hi MDG,

Variations come from things like product, process, sampling method, analytical method, actual requirements,  eg -

(1) Product / process being sampled ?, eg solid, liquid, pipe, tank, conveyor belt, lumps, powder etc etc

(2) methodology used to obtain the average numbers, eg being derived from plate count or MPN values ? units cfu/g or MPN/g or ?

(3) meaning, eg is (a) a pair of samples taken at each time and duplicate plates for each sample averaged or (b) one sample is taken and duplicate plates are averaged, or (c) something else ?

(4) yr specification for max coliform ?

If the answer to (4) is 1000 MPN/gram then the variations observed are presumably irrelevant unless you are interested in statistics.

In comparison if the spec. is max 10 MPN/gram the variation may be of immediate interest.

Note that the analytical confidence intervals associated with individual microbiological measurements are often large, eg +/- 50 %.

Kind Regards,

Charles.C

### #3 MDG

MDG

• IFSQN Member
• 98 posts
• 5 thanks
1
Neutral

• India

• Gender:Male
• Location:India

Posted 19 August 2017 - 05:47 PM

Dear Charles,

1) Process samples are drawn and sample is semi solid : ice cream

2) It's derived from Plate count (cfu/g)

3) One sample is taken and duplicate plate is averaged

4) 20

### #4 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 19 August 2017 - 07:12 PM

Dear Charles,

1) Process samples are drawn and sample is semi solid : ice cream

2) It's derived from Plate count (cfu/g)

3) One sample is taken and duplicate plate is averaged

4) 20

Hi MDG,

I assume the "20" = max. 20 cfu/g

I assume the lab method involves diluting a sample 9:1 prior to analysing, eg 25g diluted to 250 ml solution. Then plating 1ml.

If so this will unfortunately IMO make yr results too inaccurate for reliable comparison to yr limit. i think yr lab will understand the reasons. An alternative methodology is required.

If yr Procedure is different to the above please inform details, preferably with one (any) example of the calculation used to give one (any except zero) plate count value. (it is [mathematically] questionable to report  zero for a plate count result if above context).

Note - At least 2 samples/event would IMO be preferable. Especially if non-homogeneity is suspected. But i appreciate yr lab might be overloaded.

Kind Regards,

Charles.C

### #5 MDG

MDG

• IFSQN Member
• 98 posts
• 5 thanks
1
Neutral

• India

• Gender:Male
• Location:India

Posted 20 August 2017 - 03:05 PM

Dear Charles,

Yes, we follow the lab method involves diluting a sample 9:1 prior to analysing, e.g. 25g diluted to 250 ml solution, as MPN method is not feasible due to over load of samples.

Request for some help.

### #6 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 21 August 2017 - 06:27 AM

Dear Charles,

Yes, we follow the lab method involves diluting a sample 9:1 prior to analysing, e.g. 25g diluted to 250 ml solution, as MPN method is not feasible due to over load of samples.

Request for some help.

Hi MDG,

There are some specialised methods for estimation of low levels of coliform in food/dairy products such as membrane filters, AOAC 996.02 and others. Can google “estimation coliform in ice-cream” if interested.

But offhand the only “simple” procedure that I know is via MPN.

PS - one suggestion which might help a little is to try diluting 5:1 instead of 9:1,  if the solution is still practical to use.

Kind Regards,

Charles.C

### #7 FurFarmandFork

FurFarmandFork

Food Safety Consultant, Production Supervisor

• IFSQN Fellow
• 1,264 posts
• 584 thanks
185
Excellent

• United States

• Gender:Male
• Location:Oregon, USA

Posted 21 August 2017 - 10:43 PM

Note that the analytical confidence intervals associated with individual microbiological measurements are often large, eg +/- 50 %.

SUPER ditto. Heck, the AOAC proficiency testing results often vary by more than 1 log when you get out to the confidence limits for each sample.

3m has a short whitepaper discussing confidence limits (and promoting their product). Keep in mind that this is an ad for petrifilm, but does a good job of showing the variation at the lower detection threshold of the method.

Austin Bouck
Owner/Consultant at Fur, Farm, and Fork.
Consulting for companies needing effective, lean food safety systems and solutions.

Subscribe to the blog at furfarmandfork.com for food safety research, insights, and analysis.

### #8 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 23 August 2017 - 06:36 AM

Hi MDG,

I can add two more comments which may relate to yr "project".

(1) I anticipate you are using VRBA agar for  the coliform plate counts. IIRC ( I only use MPN myself) it is important to verify that the colonies counted actually are coliform since other, non-coliform, genera can sometimes grow on this agar.

(2) If you must use a plate-based procedure, IMO you need to estimate the accuracy of an average count such as you have associated with the various times. This requires taking, for example, at least 3 samples (not plates) at each time. This enables an approximate estimation of the error in the average value so that you can have more confidence in predicting whether the variations in counts as reported in yr OP are genuine changes or just due to random sampling/analytical variations of samples which may be considered as originating from the "same" batch.

The obvious difficulty with (1,2) above is that the workload is increased so that the overall time/labour may be comparable with a MPN-based procedure (although even the latter may require consideration of point (2) above since MPN confidence intervals are also large despite the procedure's increased sensitivity).

Kind Regards,

Charles.C

### #9 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 24 August 2017 - 06:28 AM

SUPER ditto. Heck, the AOAC proficiency testing results often vary by more than 1 log when you get out to the confidence limits for each sample.

3m has a short whitepaper discussing confidence limits (and promoting their product). Keep in mind that this is an ad for petrifilm, but does a good job of showing the variation at the lower detection threshold of the method.

Hi 3F,

I note that there is no mention of -

(a) the possibility of using >= 5 MPN tubes per dilution instead of 3 (or even >=5 tubes at one dilution).

(b) Detection limits (ie "somewhat more sensitive " ), eg see where the blue dots stop.

MPN procedures unquestionably have their own limitations but they can provide estimates for some situations where plate counts are simply unable to go.

Kind Regards,

Charles.C

### #10 MDG

MDG

• IFSQN Member
• 98 posts
• 5 thanks
1
Neutral

• India

• Gender:Male
• Location:India

Posted 25 August 2017 - 04:18 PM

Thank you for valuable guidance.

### #11 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 26 August 2017 - 05:40 AM   Best Answer

Hi MDG,

(1) I presume yr stated limit of coliform <10cfu/g  refers to the finished product and therefore, in practice, an accumulated "lot" of the product. It is unclear what/where yr own sampling is being carried out.

If this limit is associated with some particular Standard, it is common to find that the Standard also defines the sampling/analytical procedure used to demonstrate compliance.

IMEX, it is more common to find micro. lot acceptance requirements stated in a nmcM format. For example I noted an ice cream standard elsewhere of the form 5/10/2/100. A sole criterion such as <10cfu/g is Ok as a guideline target but sort of statistically limited from an acceptance POV unless it refers to an average result based on a defined number of samples.

(2a) There is an alternative way to utilise data from yr OP via either basic statistics or a process POV.

If you assume that yr set of data are taken from a process "in control", the average/std.dev. of the data can be used to predict the probability that, for the subsequently accumulated batch, any random sample from the batch will yield a result  <10cfu/g (ie compliance with spec) (assuming same analytical procedure).  Conversely, you can estimate the necessary target average required if the first calculation gives too low a probability of compliance. This uses basic statistics formulae and should be quite easy to apply.

As an extension of (2a) you could consider implementing a process control procedure (again basic textbook stuff) -

(2b) Assuming that you set a target average coliform process level to be such that, theoretically, any random sample of the subsequent accumulated lot will  "likely" match yr OP-spec (ie <10cfu/g), yr timed sampling data can be used to test whether yr process is in suitable "statistical control" to consistently match such a requirement.

This means that you set yr target process average level of coliform at a certain value, say X (below 10cfu/g), such that a random sample from the resulting batch will have, for example, an approx 99% probability of giving a result of < 10cfu/g (3 sigma outer control limit).

This requires (a) defining X from an estimated Process std. deviation (via sampling data)  and then (b) using timed sampling data to test whether the process data conforms to the required  control limit. Usually this procedure requires at least 2 samples being analysed at each designated time.

The above set-up is fully described in any textbook on process statistical control. Once you've done the initial calculations, you just plot the points with time.

One caveat to the above is that inaccurate analytical data may generate large standard deviations so that the adequacy of process control becomes difficult to evaluate.

Kind Regards,

Charles.C

 Thanked by 1 Member:

### #12 MDG

MDG

• IFSQN Member
• 98 posts
• 5 thanks
1
Neutral

• India

• Gender:Male
• Location:India

Posted 27 August 2017 - 03:08 PM

Dear Charles,

Thank you, We will work on POV to get the exact deviation  and  will update you on the same in 15 days.

### #13 Charles.C

Charles.C

• IFSQN Moderator
• 18,249 posts
• 5115 thanks
1,118
Excellent

• Earth

• Gender:Male
• Interests:SF
TV
Movies

Posted 28 August 2017 - 03:44 AM

Hi MDG,

As noted in a previous post, assuming the changes described below still permit satisfactory use of yr plating/evaluating procedure, some improvement of sensitivity with the plating method is possible, eg –

Assuming sample size is 25g, homogenising the sample with 100g of the usual diluent will give a solution of dilution ratio 1:5, ie 1gram (~1ml) of homogenate will contain 0.2 gram of sample.

By placing 1ml of this homogenate on each of 2 agar plates (ie total 0.4g) a lower limit of detection of 2.5 cells per gram of original sample can be reached. (This compares to (the usual) limit of detection of <10cfu/gram).

This “variation” has been published as an alternative to MPN for estimating E.coli in meat (using membrane filter/larger than normal size agar plates) but it may/may not be practical with yr product / Coliform culture plates.

The only equipment difference to yr current method would presumably be to use 2 plates instead of one. Plus some time. But the homogenate may be too viscous to use.

Just a suggestion.

Kind Regards,

Charles.C

 Thanked by 1 Member:

#### 0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users