Device Classification Under EU MDR: A Practical Guide to Annex VIII

Why classification deserves careful attention

Your class sets off a chain of decisions — which conformity route you follow, whether you need a Notified Body, how much clinical data you have to build, what post-market surveillance looks like, and how long your certificate lasts. Getting this right early saves a lot of rework later. Getting it wrong and discovering the error during a Notified Body audit is one of the more expensive compliance mistakes a manufacturer can make.

MDR introduced a new class structure — Class I, IIa, IIb, and III — that builds on MDD, but changed some rules that pushed devices into higher classes than they were under the old framework. Reusable surgical instruments that were Class I under MDD are now Class Is or higher under MDR. Active implantable devices that were Class III remain Class III but with tighter clinical requirements. Many devices that manufacturers assumed were staying at the same class shifted upward.

How the Annex VIII rules work

Annex VIII organises classification into implementing rules and classification rules. The implementing rules (Rules 1–4) set the structural logic: you apply whichever rule gives the highest classification if multiple rules apply. The classification rules (Rules 5–22) cover specific device types: non-invasive, invasive, active, special.

The process is not algorithmic in the way a flowchart might suggest. You are answering a series of questions about the device:

• What does it actually do to or in the body?
• How long is it in contact with the body — transient (under 60 minutes), short-term (up to 30 days), long-term (more than 30 days)?
• Is it invasive? If so, what kind of invasiveness — does it penetrate through a body orifice, or surgically through the skin?
• Is it active — does it depend on a source of energy other than gravity or the human body?
• Does it deliver energy, substances, or radiation to the body?
• Is it used in combination with another device?

The answers to these questions navigate you through the rules. Most errors happen because manufacturers answer them based on typical use rather than on the device's full intended purpose and all realistic modes of use.

Where classification goes wrong

The most common source of classification errors is intended purpose drift. Manufacturers often write an intended purpose that is narrower than how the device is actually being marketed or used, then classify based on that narrower purpose. Notified Bodies read the IFU, the website, and the sales materials. If those imply uses not covered by the stated intended purpose, the classification analysis is undermined.

Rule 11 — which covers software — is another frequent trouble spot. Under MDR Rule 11, software that influences clinical decisions affecting individual patients with potential to cause death or serious deterioration of health is Class III. Software that influences decisions in other ways or monitors physiological processes and could cause serious deterioration is Class IIb. The downgrade criteria are strictly defined. Many manufacturers initially place software too low and have to revise upward after the Notified Body review.

A subtler issue affects combination products — devices that incorporate a medicinal substance or use an ancillary medicinal substance as an integral part. Rule 14 pushes these to Class III. Teams managing a drug-device combination often underestimate this early and are surprised when they realise what Class III clinical evidence requirements look like.

Documenting your classification

Your classification rationale belongs in your technical documentation (Annex II or Annex III depending on your class). It needs to be more than a statement of the outcome — it needs to walk through the applicable rules, explain why other rules were considered and excluded, and connect the analysis to your intended purpose.

A useful structure:

1. State the intended purpose precisely — aligned with your IFU
2. Apply implementing rules 1–4 to establish the framework
3. Work through each relevant classification rule, noting why it applies or does not
4. Identify the highest applicable class
5. Cross-reference to the intended purpose and device description
6. Date and version-control the analysis — it needs to be updated if intended purpose changes

If your classification changes between your MDD and MDR technical files, document the reasons explicitly. Notified Bodies will ask about this, and "we reviewed it more carefully under MDR" is a legitimate answer — as long as the analysis supports it.

A note on reclassification and borderline classification

If you disagree with your Notified Body's classification opinion, you can escalate to the relevant national competent authority. This is not a common path, but it exists. The MDCG has also published classification guidance for specific device types (including MDCG 2021-24 on software classification) that Notified Bodies are expected to follow, though they are not bound by it.

For borderline classification cases — particularly around whether a device is Class IIa or IIb — it is worth getting an early opinion from your Notified Body before finalising your technical documentation. Many Notified Bodies offer a preliminary review service for this. It costs time and money, but it is far cheaper than rewriting your clinical evaluation after a formal technical review rejects your classification.