I cannot say what the background is, but I can show my opinion on what it means in practice as I've used this approach with several clients.
Take care because this is a long read with several different concepts being woven together.
First, there's the essential requirement that a clinical evaluation is required for all devices. This makes sense.
Annex X mentions:
As a general rule, confirmation of conformity with the requirements concerning the characteristics and performances referred to in Sections 1 and 3 of Annex I, under the normal conditions of use of the device, and the evaluation of the side-effects and of the acceptability of the benefit/risk ratio referred to in Section 6 of Annex I, must be based on clinical data. The evaluation of this data, hereinafter referred to as "clinical evaluation", where appropriate taking account of any relevant harmonized standards, must follow a defined and methodologically sound procedure based on...
Also, clinical data means:
clinical data" means the safety and/or performance information that is generated from the use of a device. Clinical data are sourced from:
– clinical investigation(s) of the device concerned; or
– clinical investigation(s) or other studies reported in the scientific literature, of a similar device for which equivalence to the device in question can be demonstrated; or
– published and/or unpublished reports on other clinical experience of either the device in question or a similar device for which equivalence to the device in question can be demonstrated;
The mentioned requirement:
Where demonstration of conformity with essential requirements based on clinical data is not deemed appropriate, adequate justification for any such exclusion has to be given based on risk management output and under consideration of the specifics of the device/body interaction, the clinical performances intended and the claims of the manufacturer. Adequacy of demonstration of conformity with the essential requirements by performance evaluation, bench testing and pre-clinical evaluation alone has to be duly substantiated.
is related to the exceptions to the "as the general rule".
What does this mean in practice?
Section 1 of Annex I is usually seen as the ER related to safety and to the acceptability of the benefit/risk profile
Sections 3 of Annex I is usually seen as the ER related to performance
Sections 6 of Annex I is usually seen as ER the related to acceptability of undesirable side-effects
(you can clearly see this in
MEDDEV 2.7.1 Rev 4).
Although the MDD itself does mention benefit a few times, including the general requirement on benefits outweighing risks, it does not have a definition of benefit nor a definition of clinical performance. This is one of the problems that lead to, historically, CE marking being focused in device performance and not on patient benefit.
MEDDEV 2.7.1 Rev 4 does have a definition of "clinical performance" that also includes a "definition" of benefit.
Clinical performance: behaviour of a medical device or response of the subject(s) to that medical device in relation to its intended use, when correctly applied to appropriate subject(s). [EN ISO 14155:2011]
From the definition above, "behaviour of a medical device" related to what the device do (performance) and "response of the subject(s) to that medical device" is what happens to the patient after the device is used (benefit).
How to we verify that? We can generically separate things this way (although there my some exceptions):
To verify device performance: I may not need clinical data (I would say that most of the time you do not need clinical data to verify device performance) because I need to know what the device do (independently of what happens to the patient)
To verify patient benefit: I obviously need clinical data, because I need to know what happens to the patient, not the device
So, going back to the OP:
- If my device "clinical performance" is only related to the device performance part of the definition, in principle I would not need clinical data to verify it (an thus show compliance with ER 3)
- If my device clinical performance includes patient benefit, I do need clinical data to verify it.
One example that I've been working for some time with a client: Breast implants.
Clinical performance of a breast implant (situation 1) is historically seen as related to chest-size change and bra cup-size change, patient satisfaction, and quality of life (QOL).
All of these are related to what happens to the patient (in the first two, it's what physically happens to the patient, and the last two it's what psychologically happens to the patient). All of these need clinical data.
Now, let's say that hypothetically (situation 2) that instead of these 4 aspects of clinical performance, the only relevant aspect was the implant size (which is a characteristic of the device) - if the implant size is within x cm +- y, it's ok. I don't need clinical data to evaluate the implant size, I only need to bench measure it.
Now let's say that, instead of the 4, we have, for clinical performance (situation 3), the hypothetical implant size, but also have patient satisfaction and quality of life. So here the clinical performance would be a mix of device performance and patient benefit.
In these cases, clinical data would be needed in Situation 1 and 3, but not in 2.
This is a very simplistic view and example, and the problem is that it deals with only ER 3, not 1 and 6.
Going back to those:
Section 1 of Annex I is usually seen as the ER related to safety and to the acceptability of the benefit/risk profile
The ER related to safety includes clinical safety (defined as freedom from unacceptable clinical risk) and general safety (all other safety, such as electrical, mechanical safety, and also including aspects such as usability (use error safety)). In the case of non-clinical safety, we can usually use things like standards to show compliance. In the case of clinical safety, we need to verify what are the safety risks. If safety risks can be verified by means other than clinical data (usually in the case we can justify that for example compliance with standards alone is sufficient for clinical safety), then you can use this as part of the justification of non using clinical data
The acceptability of the benefit/risk profile depends on what the benefit of the device is - is it related to device performance only or it does include patient benefit?
This is also the reason the "Where demonstration of conformity with essential requirements based on clinical data is not deemed appropriate, adequate justification for any such exclusion has to be given based on risk management output..."
Section 3 of Annex I is usually seen as the ER related to performance
This include clinical performance and general device performance.
Clinical performance, as mentioned above, may include device performance or patient benefit.
One important note: generally, we cannot use standards for clinical OR device performance. There are some product standards that may define the full range of performance of simpler devices, but usually what we have is something like 60601, which deals with only a fraction of performance when it defines requirements for what it call "essential performance".
Sections 6 of Annex I is usually seen as ER the related to acceptability of undesirable side-effects
This also relates to something that happens on the patient, not device performance. So this is also needs clinical data. However, it may be the case that the side effects are negligible or are already historically known and the residual risk is acceptable, and in this case we usually can use a literature route.
So, for a summary of my opinion, you can justify not using clinical data if (and I may be missing some more requirements here):
- device safety can be evaluated by performance evaluation, bench testing and pre-clinical evaluation
- clinical safety can be evaluated by means other than clinical data
- benefit/risk profile does not include a benefit to patient component
- clinical performance does not include a benefit to patient component and thus can be evaluated by performance evaluation, bench testing and pre-clinical evaluation
- side-effects are either negligible or historically known, the residual risk is acceptable and thus you do not need clinical data
- your risk management shows that everything above is true
- you do not claim anything that requires clinical data