How to determine sample size to be confident of result?

Hi,

I like to know how to sample finished product or ingredient so the sample to be tested is representative of the product as a whole so it will increase confidence in subsequent test result. This is need for allergen validation of finished products and to verify incoming ingredients are gluten-free.

Thanks

Konesh

A composite sample of all product is a representative of the products. Advice is to take samples from start, middle and end of production.

A positive will inform detailed testing.

Thanks

Pinks. 

Take a look at the BSi 6001 for the standards developing sampling plans.

George

A composite sample of all product is a representative of the products. Advice is to take samples from start, middle and end of production.

A positive will inform detailed testing.

 

Thanks

 

Pinks. 

We also did composite sampling. You can also ask your allergen test kit supplier if they have different recommendations based on your facilities or the nature of your product.

Regards,

Sigrid

Thanks everyone for your valuable imput.

Konesh

Dear konesh,

Yr question as far as "representative" is involved is primarily statistical. Detection is a different aspect

And statistically similar to sampling for Salmonella if you are referring to an isolated lot of individual packages of product derived from  (possibly) various sources.

The typical statistical requirement is one of determining  a sample size such that one can state that there is, for example, a 95% probability (ie confidence level)  that the level of contamination in the lot is less than X%. Numerous publications detail the solution for this, eg the classic publication Microorganisms in Food Vol.1(?)  from ICMSF (later updated in vol8 (?) which latter may be available on-line as google book).

In practice, some interpretations arbitrarily set a standard such that compliance (eg null detection result for 15 random samples) allows one to "claim" that the lot is "free" of Salmonella. Strictly speaking, this is of limited statistical meaning of course. :smile:

Related extracts from some publications may also exist on this forum if you do a little searching.

Rgds / Charles.C

Hi,

I agree with Charles C., salmonella sampling plans might be helpfull (see BAM or ICMSF). But here you have to keep in mind that the statistic behind that is, that you test on "absent" i.e. you can pool samples because you are looking for a self propagating (living) system. Dilution below LOD is no a question in that case (with some limits), but with allergens as analyte that is possible if sample is not homogeneous contaminated.

Rgds

moskito