The MDR has been applicable for some time and the IVDR just became applicable.

The MDCG has, since both regulations entered into force, issued more guidance documents for the MDR and the IVDR in a few years than were ever issued for the old directives in several decades. Sounds like progress! High fives all around in the MDCG. Look at us crank out guidance by the truckload for all mankind!

Also, the effect of using regulations instead of directives and a more centralised governance structure like the MDCG would lead to a more unitary interpretation than before. And MDCG guidance helps for unitary interpretation! More progress!

But as always, it doesn’t matter so much what you have but it’s what you do with it that counts. I am seeing a disturbing and rather unlawful development in the way the member states in the MDCG impose guidance on stakeholders as binding rules, regardless of the huge disclaimer on every MDCG guidance that ‘any views expressed in this document are not legally binding and only the Court of Justice of the European Union can give binding interpretations of Union law’ (this is actually a legally incorrect disclaimer by the way, because all Union courts and competent authorities can give binding interpretations of Union law; the Court of Justice is the highest appeal instance in the Union for binding interpretations of Union law).

‘Binding’ MDCG guidance

The development that I see in relation to guidance is the MDCG’s tendency to legislate by means of guidance, and the MDCG member member states’ tendency to interpret guidance as ‘mandatory’, which means that they impose it on notified bodies and industry as ‘mandatory’. This, in turn, leads notified bodies to raise non-conformities against guidance itself, rather than against binding regulatory requirements or to flatout ever want to discuss an interpretation set out in MDCG guidance as an option and not a mandatory solution.

As a lawyer that hurts my constitutional feelings because it’s a constitutional no-no. It’s not allowed. If you want to make binding rules, you have to follow the procedures that we have for this in a democratic legal operating system like the EU. That’s rule of law 101. If you don’t, you add to the – in this case justified – criticism that Brussels imposes rules undemocratically. If the mandatory rules are unclear, make better rules and not guidance that is imposed as if it is binding.

European Court of Justice says guidance itself is NEVER binding

Do you know who else’s constitutional feelings are hurt by this? The European Court of Justice in Luxemburg, the end boss for interpretation of binding law in the Union.

The CJEU has ruled on several occasions in no uncertain terms that guidance on the interpretation of EU law as such cannot be a source of binding law, but can only assist in its interpretation. In each of these cases it concerned a EU institution or a Member State authority seeking to rely directly on EU issued guidance for interpretation of a legal standard, which prompted the CJEU to rule in the famous Chemische Fabrik Kreussler case:

“23. In that regard, it must be stated that, in itself, that guidance document drawn up by the Commission’s services, which is not one of the legal acts of the European Union referred to in Article 288 TFEU, cannot be of a legally binding nature or enforceable against individuals.

24. That is moreover apparent from that document, which states that it is not legally binding since only the Court of Justice has the jurisdiction to give a binding interpretation of European Union law.

25. As stated by the guidance document on the demarcation between the Cosmetic Products Directive and the Medicinal Products Directive in its introduction, the fact remains that, inasmuch as that document was drawn up by group of experts from the national authorities, the Commission’s services and professional associations from industry, it may provide useful information for the interpretation of the relevant provisions of European Union law and therefore contribute to ensuring that they are applied uniformly.

26. Consequently, the national court may, in order to apply the term ‘pharmacological action’ within the meaning of Article 1(2)(b) of Directive 2001/83, take account of that document. In doing so, it must nevertheless ensure that the interpretation thus derived was derived in a manner consistent with the criteria laid down by the case-law relating to the interpretation of European Union legal acts, including those concerning the division of jurisdiction between the national courts and the Court in the context of preliminary ruling proceedings.”

Chemische Fabrik Kreussler, points 23-26

This means that:

– The MDCG should/shall (pun intended, see below) not write guidance in the expectation of it being followed to the letter;

⁃ Member States should/shall (same pun) not impose mandatory literal use of guidance on notified bodies;

⁃ Member States themselves should/shall (still the same pun) refer to guidance as a literal source of binding rules and impose them on stakeholders as law; and

⁃ Notified bodies cannot raise non-conformities directly against guidance only because guidance is not a binding rule.

A proper and legally correct finding of non-conformity cites the legal basis in the law (an article and subclause (if any) – not a recital), case law (because mandatory) and perhaps harmonized standards after a discussion why the solution offered by the manufacturer is not as good as the one in the harmonized standard. But the finding never refers only to non-binding guidance just because it’s practical for the competent authority or notified body. Non-binding guidance is, by its very nature, not a ‘requirement’ that a non-conformity can be raised against. ISO 17021 is pretty clear about that for notified bodies. It could only be seen as a requirement if there is a legal basis for it to be a requirement, which there is not according to the CJEU. Because if guidance it’s not as such a requirement, because it’s not binding.

One of the areas in which I think we are going to see more of this in the short term is the area of borderline products, for which the MDCG has just produced a shiny new MDCG guidance (MDCG 2022-5) that took years for member states to align on, so they will very likely try to impose this to the letter, regardless of the CJEU’s case law that says they don’t need to (e.g. the Lycocentre case). But that would be illegal, as was confirmed by the CJEU in – ironically – a case about guidance about borderline products and specifically about the interpretation of the concept of pharmacological action (Chemische Fabrik Kreussler, points 29-30). Rather, for each product the borderline needs to be substantiated, because

“products containing a substance having a physiological effect cannot automatically be classified as medicinal products by function unless the competent administration has made an assessment, with due diligence, of each product individually, taking account, in particular, of that product’s specific pharmacological, immunological or metabolic properties, to the extent to which they can be established in the present state of scientific knowledge.”

Hecht Pharma, point 40

Using guidance as law definitely will also not pass the Commission’s own Better Regulation criteria. I certainly appreciate the need and wish of the MDCG (and the Commission, which actually chairs the MDCG) to operationalise the MDR and IVDR efficiently. I really do. But what I do not appreciate as a lawyer is to promote guidance to mandatory rules to compensate for unclear legislation.


But Erik, you are such a hair-splitting lawyer you may say. The MDCG guidances never contain anything but a fair reflection and practical operationalisation of vague concepts in the law. Well, that’s actually precisely the problem. Our legislation in the MDR and IVDR is often not specific enough because it was designed as a not necessarily complete framework that was good enough to start with and would subsequently be filled in by means of delegated and implementing acts. But making it more specific by means of guidance risks that requirements are provided in guidance that are not actually required by the law, which is exactly what the CJEU rules is unconstitutional because we have democratic legislative procedures to create binding legislation. This is exacerbated by absence of proper legal review / appeal options against notified body decisions, which is a more general legal problem perpetuated in the MDR and IVDR.

Example: I give you the double importer check in MDCG 2021-27 (question 7), which is literally nowhere to be found in the law, but is put in the guidance as something that importers should do. I raised this as imposing an additonal requirement without legal basis in a previous blog and subsequently the Commission confirms to me alone that it is in fact optional, not mandatory. Yet, this is not how the guidance puts it and how notified bodies may enforce this guidance against manufacturers. Based on how the guidance is worded, a competent authority may well use this as a basis for enforcement. This puts the finger exactly on the problem that MDCG guidance is not necessarily what the law prescribes.

Another example is the overly legalistic and frankly illogical interpretation interpretation of rule 8 regarding accessories to active implants in MDGC 2021-24 on classification. Because the AIMDD did not contain a separate concept of accessory, rule 8 of Annex VIII now states

“All implantable devices and long-term surgically invasive devices are classified as class IIb unless they: 


— are active implantable devices or their accessories, in which cases they are classified as class III;” 

Rule 8, Annex VIII MDR

Admittedly, this is not the best legal drafting ever seen in a regulation. However, this is a rule about implantable devices, in the implantable devices chapter of the classification annex of the MDR, and it would be entirely logical that it applies only to implantable accessories. Why? Because the implementing rules in Annex VIII that explain the logic behind classification state that

“Accessories for a medical device and for a product listed in Annex XVI shall be classified in their own right separately from the device with which they are used.”

Implementing rule 3.2, Annex VIII MDR

In other words, non-implantable accessories should be classified in their own right, and not automatically in the same risk class as the active implantable device itself. Yet, MDCG 2021-24 states that in scope of this rule are:

“Accessories to active implantable devices (with or without contact to the heart), be it implantable or non-implantable active or not: 
– torque wrench for implantable pulse generator / implantable cardioverter defibrillator
– cables for programmer / pacing system analyser
– magnet for Implantable Pulse Generator / Implantable Cardioverter Generator
– programmer or an external transmitter intended for activating or controlling the implantable part of the device
– implantable pacemaker leads”

MDCG 2021-24, p. 38

Implantable pacemaker leads covered by a classification rule about implantable devices? Totally agree. You know why? Because they are implantable so it’s logical that they are covered by a rule on implantable devices.

Magnet or torque wrench? Baloney! These would be class I devices if classified in their own right, as they should be according to MDR classification logic under implementing rule 3.2.

The interpretation is not practical, it was not even initially chosen by notified bodies and is now imposed on notified bodies, some of which had to change their interpretation affecting certificates already issued. Does that serve patient safety better? Does it improve the performance of the devices concerned?

To put it euphemistically: I am not convinced at all that this is an improvement. So now we have class III magnets and torque wrenches. We get to do a PSUR for a torque wrench or a magnet. At least we can consult an expert panel with a proposal for clinical investigation (placebo magnets anyone? that will be fun in the cath lab). And we get to justify why there is no need clinical investigation for a magnet or torque wrench. It is kind of weird that you don’t need clinical investigation for an implantable wedge, screw, plate, etc., but you have to justify it for a new magnet or torque wrench. Again, is this logical? No. Is it conducive to device performance or patent safety? I don’t think so. Is it overly legalistic and does it create more red tape? Absolutely. And only because the notified bodies are ordered to apply this guidance to the letter, which – according to the case law of the CJEU – is blatantly unlawful.

And it’s not that the law cannot be amended to be more specific or precise: the MDR and IVDR are riddled with delegated responsibility for the Commission to propose amendments to make the law more practical or adapt it to current state of affairs by means of delegated and implementing acts. Also, when needed the EU has shown that it is willing to amend medical devices regulation even by corrigendum (also doubtful from a constitutional perspective, but no one was going to complain about it this time).

Should / shall

The frequent use of the word ‘should’ in MDCG guidance denotes that at least the good people at the Commission legal service know better than to say ‘shall’ in the guidance – they’re lawyers too, and very good ones at that. But they cannot help the MDCG enforcing should like it says shall. Maybe they should have a chat with the MDCG to explain this at some point… and then also rewrite the disclaimer for the MDCG guidances.

It is kind of cynical to demand literal application of guidance (“shall”) with a disclaimer on the front that says it’s not binding and contains optional rather than mandatory regulatory requirements and write in the guidance that you should (which is really different from ‘shall’), wouldn’t you say? It is really falling in the Platonic trap described in Plato’s Phaedrus dialogue in which Socrates warns against believing that something is true just because it was written down by someone thinking he knows what he’s talking about (Plato, Phaedrus, 275a-c). And this is the CJEU’s point: you may only assume this in case of actual law applied to the actual situation in view of an actual analysis of this specific product and all the other positions, for example those stated in abstracto in guidance, are always up for debate:

“the competent administration has made an assessment, with due diligence, of each product individually, taking account, in particular, of that product’s specific pharmacological, immunological or metabolic properties, to the extent to which they can be established in the present state of scientific knowledge”

(Chemische Fabrik Kreussler, point 33)


But if you are going to treat guidance as binding rules / pseudo legislation, then some things should at least be provided which are baseline requirements for real legislation (so sort of best practice for binding rule making you could say):

  • a transitional period for the new rule or interpretation of a rule to apply; and
  • a process for quick amendment of guidance, where mistakes or other issues are identified (such as the importer double check non-requirement)

At present all MDCG guidance is always immediately applicable without any transition, which is totally frowned upon in case of new rules imposed in legislation. There are examples of MEDDEV guidance for which notified bodies were allowed to use a transitional period, so why not for MDCG guidance?

Well, I can think of one reason: because every MDCG guidance ever issued was already (way) too late for the transitional regime to have practical use and time is of the essence for the MDCG once they have finally managed to agree on a new guidance document, which may take years in some cases. Still, that is not an excuse for rule making that does not comply with basic principles of rule making in a democratic society.

The Platonic trap

So … let’s not fall in the Platonic trap of the MDCG guidances together, and assume that they are wisdom just because they were written down by other people and to believe that they are law even if enacted outside of the correct process for enacting binding rules. This wisdom is thousands of years old.

And moreover the CJEU says you don’t have to! Competent authorities and notified bodies owe stakeholders better reasoned decisions and non-conformities than basing them on a mere reference to guidance, even if they are in a hurry or if it’s not that convenient.

The MDCG owes us better quality guidance that is actually guiding rather than mandatory because it’s unlawful. And guidance that is actually guiding to solutions, not imposing them to the letter in a legalistic one size fits all. Not even harmonised standards and common specifications do this and they actually do have a legal basis for providing presumption of conformity.

Rather, let’s be logical and practical with patient safety in mind. Let’s use our thinking brains and not be overly legalistic, or – if we want to be – let’s have a proper legal basis for it. It’s not like we don’t have enough issues with the MDR and the IVDR.

UPDATE 6 June 2022

Silly me – I overlooked that even the MDCG itself has stated in MDCG 2019-6 that in relation to the use of guidance in accordance with Annex VII, 4.5.1 (see also Annex VII, 1.6.2 with a similar requirement to ‘take guidance into consideration’) that:

“For instance, in order to assess if the solutions adopted by the manufacturer are state of the art and in line with expectations, the CAB need to use the available guidance documents and standards. It should be noted that non-conformities will not be raised against standards or guidance but need to be phrased against legal requirements.”

MDCG 2019-6 Rev. 3 Questions and answers: Requirements relating to notified bodies, IV.11

Anyway, this underlines and confirms everything said above in this post. Except that it makes it even more problematic that notified bodies would be instructed to treat guidance as law and competent authorities refuse to discuss guidance as guidance.