Number of Laboratories required to participate in ILCP

[Prakash Gupta]

Happy holi to All

For the ILCP (Inter-Laboratory Comparison Programme), how many minimum number of laboratory is required to participate to get ISO 17025.

 

[dv8shane]

Happy holi to All

For the ILCP (Inter-Laboratory Comparison Programme), how many minimum number of laboratory is required to participate to get ISO 17025.

To my knowledge the minimum number is 2.

You should be able to do a combined ILC and proficiency test if you can get your national lab and/or your designate. to agree and participate.

You would take the readings of the standard to be calibrated prior to sending it out.
You ask the organization calibrating the standard to with hold their findings until you have received the item back, taken another set of readings and sent both sets of readings to them.

The accepted basis for comparison of 2 or more laboratories utilizes En or normalized error.

 

[zamclachia]

My understanding is that a minimum of 8 laboratories need to partisipate. Less than 8 and the data produced is statistically invalid. There is an ISO standard covering this. Is there anyone out there that can give a reference for the standard?

 

[dv8shane]

My understanding is that a minimum of 8 laboratories need to partisipate. Less than 8 and the data produced is statistically invalid. There is an ISO standard covering this. Is there anyone out there that can give a reference for the standard?

My country's national lab and accrediting service has this on file.
http://www.nrc-cnrc.gc.ca/obj/inms-ienm/doc/clas-clas/clas_requirements_document_07.pdf

Having authored and pivoted several ILC's I do not understand where the number 8 comes from. It is a comparison between 2 laboratories especially if you are trying to prove proficiency. You can have as many participants as you want and each is compared to all the others. I have a sample I could send you.

 

[zamclachia]

When the parameter been analysed is not well defined, a minimum of 8 results from different labs is required in order to calculate a consensus value (mean or medium depending on statistical approach used) as well as SD(Reproducibility). From this a Z score can be determined to evaluate the participants performance.
Z = (result - consensus)/SD(R)
Where Z < = 2 is acceptable, 2 > Z < 3 is questionable and Z >= 3 is unacceptable.