Sampling method that is representative of the devices supplied by that distributor

[dieperie]

Dear all,

I am trying to fulfill/address the following requirement:

Regulation (EU) 2017/745 - Article 14(2) - General obligations of distributors

"In order to meet the requirements referred to in points (a), (b) and (d) of the first subparagraph the distributor may apply a sampling method that is representative of the devices supplied by that distributor."

Does anyone in the community have suggestions ?

Kind regards,

Edgar​

 

[Mike S.]

Re: Sampling method that is representative of the devices supplied by that distributo

What you have posted is a vague requirement. Is there no further guidance anywhere?

If not, you have to decide (and perhaps defend) what you would determine to be a representative sample. I don’t know your products, lot sizes, and what kind of variation you might expect to find in a lot, or the risks associated with not finding “excessive” variation if it exists.

Maybe a discussion with your co-workers and/or customer or peer companies would give you some ideas.

 

[dieperie]

Re: Sampling method that is representative of the devices supplied by that distributo

The concern I have is with the following :

Before making a device available on the market, distributors shall verify that all of the following requirements are met:​

(a) the device has been CE marked and that the EU declaration of conformity of the device has been drawn up;​

(b) the device is accompanied by the information to be supplied by the manufacturer in accordance with Article 10(11);
(d) that, where applicable, a UDI has been assigned by the manufacturer.

One would say you verify these elements once before making available on the market. We are not debating about the Q&C, but QA aspect of product.

Methodology

The sample should be partly selective based on the factors set out above and partly nonselective, and should result in a representative range of different products being selected, without excluding the random element of sampling.

Any ideas?​

 

[Mike S.]

Re: Sampling method that is representative of the devices supplied by that distributo

Sounds like you would randomly select 1 or more Type A devices, 1 or more Type B devices, 1 or more Type C devices, etc. to satisfy the randomness requirement but also ensure that each Type of device is selected.

 

[dieperie]

Re: Sampling method that is representative of the devices supplied by that distributo

You are correct. The question is how to determine and track --> sampling plan.

 

[Statistical Steven]

Re: Sampling method that is representative of the devices supplied by that distributo

You are correct. The question is how to determine and track --> sampling plan.

The requirement seems to imply that a statistically sound sampling plan can be used to show that the requirement has been met without the need for 100% sampling. Any sampling plan should be based on the risks associated with failure to meet the acceptance criteria. In other words, you need to select a sampling plan that is commensurate with risk.