MDCG 2022-5 Borderline Products: The Mode-of-Action Question

The question "is this a medical device or a medicinal product" sounds like it should have an intuitive answer. It doesn't. Under MDR, a product is a device only if it does not achieve its principal intended action by pharmacological, immunological, or metabolic means — and the guidance manufacturers reach for to apply that test, MDCG 2022-5, gets misread constantly — usually by companies substituting intuition about how a product looks or works for the mechanism-of-action evidence the test actually requires.

The legal test: principal mode of action

MDR Article 2(1) defines a medical device as an article intended for a specific medical purpose that "does not achieve its principal intended action by pharmacological, immunological or metabolic means, in or on the human body, but which may be assisted in its function by such means." Article 1(6)(b) sharpens the test further: the principal mode of action is whatever means — pharmacological, immunological, metabolic, physical, or other — achieves the product's principal intended action, and that determination has to be objective, grounded in state-of-the-art scientific data. A product can include a substance that acts pharmacologically, immunologically, or metabolically and still be a device, provided that action is genuinely ancillary to a physical or mechanical primary effect. The moment the pharmacological, immunological, or metabolic mechanism becomes the principal action, MDR no longer applies — Directive 2001/83/EC does, and the product needs a marketing authorization, not a technical file. This is a scientific-evidence question folded into a legal one, which is exactly why it belongs inside a broader EU MDR & IVDR regulatory strategy rather than a one-off classification memo.

What MDCG 2022-5 actually adds

MDCG 2022-5 does not change the legal test — it cannot, since that test sits in the MDR text and Directive 2001/83/EC. What it adds is applied interpretation: worked examples and flowcharts covering the product categories that generate the most disputes. Herbal and substance-based products get dedicated chapters, because a substance's mechanism can shift from physical (a mechanical barrier, an osmotic or coating effect) to pharmacological (receptor binding, metabolic alteration) in ways that are not obvious from the finished product's appearance. Combinations of a device and a medicinal product get their own treatment here too — a related but distinct question from the ancillary-substance scenario under Article 117, and one worth checking against any existing combination product regulatory work before assuming the categories line up. The October 2024 revision added a section on products intended for cleaning, disinfecting, or sterilizing medical devices: a product with that sole purpose is itself a device, while a multi-purpose disinfectant falls instead under Regulation (EU) No. 528/2012 on biocidal products — a third regime the borderline analysis has to rule in or out.

• Herbal and substance-based products. The mechanism, not the source material, decides the regime — a plant-derived substance is not automatically "natural" and therefore device-eligible.
• Device–medicinal product combinations. Treated separately from Article 117's ancillary-substance pathway; MDCG 2022-5 addresses the classification question, not the conformity-assessment mechanics.
• Cleaning, disinfection, and sterilization products. Sole-purpose device-cleaning products stay under MDR; multi-purpose disinfectants shift to the biocidal products regime — added explicitly in the rev.1 update.
• Products assisted by an ancillary mechanism. A genuinely secondary pharmacological, immunological, or metabolic effect does not disqualify device status — but "ancillary" has to be demonstrated, not asserted.

The question is never what the product looks like or how it's marketed. It is what mechanism, demonstrated with current scientific evidence, actually produces the intended effect. The test MDCG 2022-5 is built to apply

Borderline classification isn't the Article 117 question

The most common conflation in this space is treating borderline classification and an MDR Article 117 assessment as the same analysis. They are sequential, not interchangeable. Borderline classification, governed by Article 2(1), Article 1(6)(b), and MDCG 2022-5, decides whether a product is a device in the first place. Article 117 — which operates by amending Directive 2001/83/EC's Annex I — only applies once that question is already settled in the device's favor: it governs a device that integrally incorporates a substance which, if used separately, would itself be a medicinal product, where that substance's action is ancillary to the device's own. A company that has already established device status under the borderline test still has to run the Article 117 analysis separately if the device contains such a substance, including the consultation procedure with a medicines competent authority or EMA that the amended Annex I requires. Getting the sequence backward — assuming Article 117 consultation settles the underlying classification, or skipping the borderline test because a product "obviously" contains a device component — is how manufacturers end up defending the wrong regulatory strategy under an information request.

None of this is exotic analysis, but it is evidence-heavy in a way a marketing-led classification memo rarely is. A defensible position needs the mechanism-of-action data lined up before a notified body or competent authority asks for it — not assembled afterward under time pressure once a submission has already been built around the wrong regime. Companies that treat the borderline question as a European regulatory strategy decision, made early and revisited whenever the product's formulation or claims change, are the ones who don't have to unwind a technical file and start over mid-review; see also how MDR classification rules apply once device status itself is settled.