6 replies to this topic

[a_andhika]

Dear All,

Today I am checking a specification from our candidate supplier. On microbiological assay, I found that the specification for Salmonella is "Negative/1500 g". As far as I am concern, the unit for salmonella would be negative/25 g. But after a quick browse, I found some supplier also mention another number, which is: "Negative/375 g". What is the real unit for Salmonella tests? And what is the story behind Negative/25 g? Thank you very much.


Regards,


Arya

PS: The material is a dairy product, whey powder.

[Charles.C]

Dear Arya,


The specific answer to yr unit question slightly depends on yr objective. With respect to the direct sample result, I suppose the nearest is cell/gram, analogous to plate count. "negative" = not detected (under the conditions of the test which is itself assumed to be "sensitive enough").
Negative (0?) cell/g obviously has various statistical objections

In the style of yr post, “Negative” usually means a direct test for Salmonella (cells) using a sample size of Xgram gave a negative result (as against positive)
= No Salmonella cells were detected in Xgram sample
= , < (1/X) cells / gram of sample (intuitive average)
(where X was 25, 375g. 1500g in yr examples)

(It doesn’t have to be so but often, 375g = 15 samples of 25g [combined together for analysis], similarly 60x25g = 1500g).

(less commonly, some people use a MPN analysis approach which would typically hv a different format, eg <3/gram etc as in one common micro. set-up)

Perhaps more significantly, a negative test result can be (loosely) interpreted as a probabilistic statement about the salmonella contamination in the lot from which the sample was taken.
Example : “a (-)ve detection result on (randomly taken) 60 samples of 25g (ie 1500g) gives an approx. 95% probability that the lot contains not more than 5% of contaminated units.”
From memory 15 x 25g implies not more than 20% of contaminated units.
(these numbers illustrate the problem of "proving" that a given lot is "salmonella free")

Some more precise explanations ( ICMSF plans ) are linked in this thread –

http://www.ifsqn.com...?showtopic=4590

Rgds / Charles.C

[a_andhika]

Dear Charles C.,

Thank you very much for your comprehensive post, as usual. I havent read your given thread entirely, so my deepest apologize if I am asking repeat questions.

It doesn’t have to be so but often, 375g = 15 samples of 25g [combined together for analysis], similarly 60x25g = 1500g.

My first question is, do the 60 samples were mixed (in respect to 'combined') altogether and then analyzed? Or it analyzed with separately?

The second question is, does it means that, the negative/1500 g was designed for a big lot size? Is there any particular sampling plan for Salmonella?

Regards,


Arya

[Charles.C]

Dear Arya,

My first question is, do the 60 samples were mixed (in respect to 'combined') altogether and then analyzed? Or it analyzed with separately?

Theoretically one could analyse every unit but obviously advantageous to combine if sensitivity maintained. From previous studies / practical reasons, USFDA / BAM procedure uses a maximum composite weight of 15 / 25g units (375g) so 60 samples would be 4 composites. Details for Salmonella are here –

http://www.cfsan.fda...ebam/bam-1.html

The second question is, does it means that, the negative/1500 g was designed for a big lot size? Is there any particular sampling plan for Salmonella?

The lot size is usually unrelated here unless very small (although in some other sampling systems it is adjusted for). Chosen sample size relates to risk evaluation / desired probability to detect / reject a contaminated lot (the first is explained in above link, second is discussed my prev.post / link - basically a bigger sample gives a better chance of finding a defect ie more reliably stating a lot is "free" of salmonella ).

Rgds / Charles.C

[a_andhika]

Dear Charles C.,

Thank you very very much. That would be sufficient for the mean time.


Regards,


Arya

[Bets]

I would follow all the FDA links. But if I recall reading somewhere 375g is the maximium amount that you would want to test as a sample size. Anything larger than that you can run into problems with the actual test. Certian comodities might overwhelm the sample. However you can test up to 1900g if you want you just have to perform 5 seperate 375g samples . Therefore you would have to list negative in 375g five seperate times.

[a_andhika]

Dear Bets,

Interesting information. I thought so, it really depends on the commodities. For the meantime, I am still using the 25 g. But just in case of emergency, I agree that max. 375 g is suitable for testing large batch in my material/products.