Regulatory Open Forum

For each device, you will identify a number of attributes. For the NBs sampling plan the important ones are device class, EMDN code, the Basic UDI-DI, IVDR code, and the IVP code. Note that footnote 6 says, "In case where more than one EMDN code applies to one device, only the most appropriate one of these EMDN codes will be assigned for sampling purposes".

Start with the device class.
For Class A there is no NB sampling because there is no NB involvement.
For Class B there is NB sampling based on the categories of devices
For Class C there is NB sampling based on generic device groups
For Class D there is no NB sampling because the NB reviews every TD at every audit

Typically, the NB will send you a form on which you will list all the devices for which you want the NB to provide a certificate. It will include all the attributes required for the NB to make the sampling plans.

The NB will make multiple sampling plans – one for each device category and one for each generic group. Each sampling plan will have a range, the list of devices in the category or the group.

Example: You have one category of devices and the range is five devices. Prior to the certificates, the NB will select one device for the sample, but they won't tell you until the last minute. You will need to have the TD available for all five devices in the category. Then during the four subsequent surveillance audits, they will pick one device at random. Since the sampling plan is a secret, you won't know which device until they tell you.

Example: You have one category of devices and the range is two devices. Prior to issuing the certificate the NB will pick one. During the surveillance audits they will keep picking between the two. Because the range is small, the NB focuses on PMS. The NB must do at least one TD audit.

Example: You have three categories of devices and the ranges are seven devices, ten devices, and twelve devices. The NB will set up a sampling plan for each category with the following constraints – there is at least one TD audit each audit (initial and four surveillance), each audit looks at least 15% of the TDs in the category, and over the five audits the NB looks at each TD at least one time.

A similar situation applies to each generic device group.

During the audits the NB will sample from each category and from each group. In addition, they will look at all Class D TDs.

For each device, you will identify a number of attributes. For the NBs sampling plan the important ones are device class, EMDN code, the Basic UDI-DI, IVDR code, and the IVP code. Note that footnote 6 says, "In case where more than one EMDN code applies to one device, only the most appropriate one of these EMDN codes will be assigned for sampling purposes".

Start with the device class.
For Class A there is no NB sampling because there is no NB involvement.
For Class B there is NB sampling based on the categories of devices
For Class C there is NB sampling based on generic device groups
For Class D there is no NB sampling because the NB reviews every TD at every audit

Typically, the NB will send you a form on which you will list all the devices for which you want the NB to provide a certificate. It will include all the attributes required for the NB to make the sampling plans.

The NB will make multiple sampling plans – one for each device category and one for each generic group. Each sampling plan will have a range, the list of devices in the category or the group.

Example: You have one category of devices and the range is five devices. Prior to the certificates, the NB will select one device for the sample, but they won't tell you until the last minute. You will need to have the TD available for all five devices in the category. Then during the four subsequent surveillance audits, they will pick one device at random. Since the sampling plan is a secret, you won't know which device until they tell you.

Example: You have one category of devices and the range is two devices. Prior to issuing the certificate the NB will pick one. During the surveillance audits they will keep picking between the two. Because the range is small, the NB focuses on PMS. The NB must do at least one TD audit.

Example: You have three categories of devices and the ranges are seven devices, ten devices, and twelve devices. The NB will set up a sampling plan for each category with the following constraints – there is at least one TD audit each audit (initial and four surveillance), each audit looks at least 15% of the TDs in the category, and over the five audits the NB looks at each TD at least one time.

A similar situation applies to each generic device group.

During the audits the NB will sample from each category and from each group. In addition, they will look at all Class D TDs.

Corey,

Happy to help.

For #1, the sentence is in section 4.1.2 Sampling during surveillance. I now think you are correct. It is at least 15% over each five-year certification cycle. I had always read it such that over the five-year cycle all TDs would have been covered.

For #2, section 4.1.1 Sampling during surveillance applies. Both indents say that prior to issuing the certificate "at least one representative device". This is not the full range.

For #3, the ambiguity is in the word sampled. In a different context, we may say all product lots are sampled at incoming inspection. That means that for all lots that come in the door, some pieces are examined in the sample. It is 100% of all lots, not 100% of all pieces. In this case it is 100% of all categories and groups, not 100% of all TDs. (For #1, I believed it was 100% of all TDs over the five-year cycle.)

For #4, you could assign multiple EMDN codes. The categories and groups combine "like" things together. The NB will use only the most appropriate one. You need to steer the NB to the smallest number so the categories and groups have a large number of members.

For the EMDN code in general, you may have multiple codes in the TD. One point, yet unanswered, is the number of EMDN codes allowed in the Eudamed device registration/UDI module. For categories, according to 3.1, they all have the same IVR codes. For groups, according to 3.2 second bullet, they all have the same EMDN code with the most appropriate IVP code.

There is one more issue that is not clear. In 4.2.2 on selecting the sample, both bullets say, "at least one representative device covered by a Basic UDI-DI". If a group or category includes more than Basic UDI-DI, it is not clear if this means to sample from the Basic UDI-DIs as well.

The good news is that, other assigning than assigning the codes, there is nothing for you to do. The NB will develop all of this behind the veil of secrecy and tell you which TDs are in the sample.

There is no bad news, because every TD is in a perpetual "audit-ready" state.

For each device, you will identify a number of attributes. For the NBs sampling plan the important ones are device class, EMDN code, the Basic UDI-DI, IVDR code, and the IVP code. Note that footnote 6 says, "In case where more than one EMDN code applies to one device, only the most appropriate one of these EMDN codes will be assigned for sampling purposes".

Start with the device class.
For Class A there is no NB sampling because there is no NB involvement.
For Class B there is NB sampling based on the categories of devices
For Class C there is NB sampling based on generic device groups
For Class D there is no NB sampling because the NB reviews every TD at every audit

Typically, the NB will send you a form on which you will list all the devices for which you want the NB to provide a certificate. It will include all the attributes required for the NB to make the sampling plans.

The NB will make multiple sampling plans – one for each device category and one for each generic group. Each sampling plan will have a range, the list of devices in the category or the group.

Example: You have one category of devices and the range is five devices. Prior to the certificates, the NB will select one device for the sample, but they won't tell you until the last minute. You will need to have the TD available for all five devices in the category. Then during the four subsequent surveillance audits, they will pick one device at random. Since the sampling plan is a secret, you won't know which device until they tell you.

Example: You have one category of devices and the range is two devices. Prior to issuing the certificate the NB will pick one. During the surveillance audits they will keep picking between the two. Because the range is small, the NB focuses on PMS. The NB must do at least one TD audit.

Example: You have three categories of devices and the ranges are seven devices, ten devices, and twelve devices. The NB will set up a sampling plan for each category with the following constraints – there is at least one TD audit each audit (initial and four surveillance), each audit looks at least 15% of the TDs in the category, and over the five audits the NB looks at each TD at least one time.

A similar situation applies to each generic device group.

During the audits the NB will sample from each category and from each group. In addition, they will look at all Class D TDs.

Corey,

Happy to help.

For #1, the sentence is in section 4.1.2 Sampling during surveillance. I now think you are correct. It is at least 15% over each five-year certification cycle. I had always read it such that over the five-year cycle all TDs would have been covered.

For #2, section 4.1.1 Sampling during surveillance applies. Both indents say that prior to issuing the certificate "at least one representative device". This is not the full range.

For #3, the ambiguity is in the word sampled. In a different context, we may say all product lots are sampled at incoming inspection. That means that for all lots that come in the door, some pieces are examined in the sample. It is 100% of all lots, not 100% of all pieces. In this case it is 100% of all categories and groups, not 100% of all TDs. (For #1, I believed it was 100% of all TDs over the five-year cycle.)

For #4, you could assign multiple EMDN codes. The categories and groups combine "like" things together. The NB will use only the most appropriate one. You need to steer the NB to the smallest number so the categories and groups have a large number of members.

For the EMDN code in general, you may have multiple codes in the TD. One point, yet unanswered, is the number of EMDN codes allowed in the Eudamed device registration/UDI module. For categories, according to 3.1, they all have the same IVR codes. For groups, according to 3.2 second bullet, they all have the same EMDN code with the most appropriate IVP code.

There is one more issue that is not clear. In 4.2.2 on selecting the sample, both bullets say, "at least one representative device covered by a Basic UDI-DI". If a group or category includes more than Basic UDI-DI, it is not clear if this means to sample from the Basic UDI-DIs as well.

The good news is that, other assigning than assigning the codes, there is nothing for you to do. The NB will develop all of this behind the veil of secrecy and tell you which TDs are in the sample.

There is no bad news, because every TD is in a perpetual "audit-ready" state.

For each device, you will identify a number of attributes. For the NBs sampling plan the important ones are device class, EMDN code, the Basic UDI-DI, IVDR code, and the IVP code. Note that footnote 6 says, "In case where more than one EMDN code applies to one device, only the most appropriate one of these EMDN codes will be assigned for sampling purposes".

Start with the device class.
For Class A there is no NB sampling because there is no NB involvement.
For Class B there is NB sampling based on the categories of devices
For Class C there is NB sampling based on generic device groups
For Class D there is no NB sampling because the NB reviews every TD at every audit

Typically, the NB will send you a form on which you will list all the devices for which you want the NB to provide a certificate. It will include all the attributes required for the NB to make the sampling plans.

The NB will make multiple sampling plans – one for each device category and one for each generic group. Each sampling plan will have a range, the list of devices in the category or the group.

Example: You have one category of devices and the range is five devices. Prior to the certificates, the NB will select one device for the sample, but they won't tell you until the last minute. You will need to have the TD available for all five devices in the category. Then during the four subsequent surveillance audits, they will pick one device at random. Since the sampling plan is a secret, you won't know which device until they tell you.

Example: You have one category of devices and the range is two devices. Prior to issuing the certificate the NB will pick one. During the surveillance audits they will keep picking between the two. Because the range is small, the NB focuses on PMS. The NB must do at least one TD audit.

Example: You have three categories of devices and the ranges are seven devices, ten devices, and twelve devices. The NB will set up a sampling plan for each category with the following constraints – there is at least one TD audit each audit (initial and four surveillance), each audit looks at least 15% of the TDs in the category, and over the five audits the NB looks at each TD at least one time.

A similar situation applies to each generic device group.

During the audits the NB will sample from each category and from each group. In addition, they will look at all Class D TDs.

I would like to add that it is essential to build well designed B-UDI Strategy.

As sampling will be performed as defined in MDR and MDCG 2019-13, will be based on the Basic UDI-DI that legal manufacturer defines.
A sampling out of the sampling is not allowed in MDR/ IVDR. This means that NB is not allowed to pick us 1 article of 10 variants included under 1(one) Basic UDI-DI. They will need to assess them all. NB need written evidence that all variants including accessories covered by the TD have been included in the TDAR. This can also have an effect on the pricing of the TDA.

References:

https://mpra.ub.uni-muenchen.de/70218/1/               

https://www.mathsisfun.com/data/sampling.html