EU MDR and IVDR in Practice: Five Case Studies in European Medical Device Market Access (2026)

Quick answer

What do five EU MDR and IVDR case studies show about European medical device market access in 2026?

Five illustrative composites show recurring failure modes under Regulation (EU) 2017/745 (MDR) and Regulation (EU) 2017/746 (IVDR): FDA 510(k) clearance does not transfer clinical or PMCF evidence for a Class IIb SaMD under Annex VIII Rule 11; RUO labelling does not shelter a Class C oncology assay–software combination; legacy MDD CER narratives fail MDR traceability for a Class III implant; a Class IIa digital therapeutic needs sequential CE marking then German DiGA (§139e SGB V) reimbursement; and for non-EU manufacturers the CE mark is midpoint—not completion—of access across forty-plus national markets, with EC REP, EUDAMED SRN, and phased national entry on the critical path.

Keywords: EU MDR case study, IVDR case study, EU MDR Class IIb SaMD, Rule 11 classification, FDA 510(k) to CE mark, Notified Body timeline, EUDAMED SRN registration, DiGA reimbursement Germany, BfArM fast-track, living Clinical Evaluation Report, PMCF plan, RUO Research Use Only trap, MDD to MDR transition, EU AI Act high-risk medical device, legacy device Class III remediation, EC REP authorised representative, European market access sequencing

Methodological Note

The five cases presented in this article are illustrative composites. They are not accounts of specific client engagements, and no case describes an identifiable manufacturer, product, or Notified Body.

Each composite is constructed from two evidence sources: (a) recurring problem patterns documented publicly by European regulatory practitioners in the 2026 interview literature, and (b) the verifiable procedural mechanics of Regulation (EU) 2017/745 (MDR), Regulation (EU) 2017/746 (IVDR), applicable MDCG guidance, and national frameworks such as Germany's §139e SGB V DiGA pathway. Timelines, classifications, and authority interactions reflect documented ranges and published procedure, not privileged information.

The composite method is used deliberately. Real conformity assessment files are confidential, and Notified Body correspondence is commercially sensitive. Composites allow the underlying failure modes to be examined in operational detail without misrepresenting any manufacturer's regulatory history. Practitioners should treat these as teaching scenarios, not precedents.

A terminological note: "route of administration" is a pharmaceutical construct. For devices, the corresponding field is the route of application or mode of patient contact; for in vitro diagnostics, it is the specimen type. Each case below uses the device-appropriate equivalent.

1. Introduction

The European Union's medical device framework is now several years past full applicability, and the pattern of failure has become legible. Practitioners working across the EU market in 2026 describe a consistent set of recurring problems: manufacturers who treated FDA clearance as a head start and discovered it was not; startups who deferred quality system investment and paid a multiple to retrofit it; legacy manufacturers whose MDD-era equivalence arguments collapsed under MDR scrutiny; and companies who obtained a CE mark and then discovered that European market access is not one market but forty.

What follows is an examination of five such patterns in operational detail. Each case sets out the manufacturer profile, engagement scope and duration, product characteristics, the specific regulatory pathway navigated, the substance of health authority and Notified Body interaction, and — most importantly — the remediation strategy that resolved the problem.

The cases are ordered to move from the most common (transatlantic SaMD transition) through to the most structurally underestimated (multi-market European access beyond the CE mark).

2. Case Study 1 — FDA-Cleared Cardiology SaMD Transitioning to EU MDR Class IIb

2.1 Client Profile

2.2 Project Scope and Duration

Engagement scope covered regulatory strategy, classification determination, ISO 13485 implementation, clinical evidence gap analysis and CER authorship, PMCF programme design, technical documentation compilation, Notified Body liaison, and EUDAMED registration.

Duration: 19 months from kick-off to CE certificate issuance. This decomposed as approximately 5 months QMS implementation and certification, 6 months clinical and technical documentation remediation, 7 months Notified Body conformity assessment inclusive of two clock-stopper cycles, and 1 month EUDAMED and launch preparation.

2.3 Product Characteristics

2.4 The Registration Process and Regulatory Strategy

Classification was the first strategic decision and the one that reframed the entire project. Under Annex VIII Rule 11, software intended to provide information used to make diagnostic or therapeutic decisions is Class IIa; it escalates to Class IIb where such decisions could cause serious deterioration of health or surgical intervention. Because the software's triage output influenced prioritisation decisions in suspected arrhythmia — where a missed detection carries risk of serious deterioration — Class IIb was the defensible position.

The manufacturer's initial assumption had been Class IIa, based on an informal reading that the software was "decision support, not diagnosis." This distinction does not exist in Rule 11. Accepting Class IIb early, rather than arguing Class IIa and being escalated by the Notified Body mid-review, preserved roughly four months.

The strategy adopted had four elements:

Conformity assessment route. Annex IX (full quality management system plus technical documentation assessment) was selected over Annex X/XI. For a software-only manufacturer with an evolving product, an Annex IX QMS certificate provides a more durable basis for subsequent change management than batch-oriented alternatives.

Notified Body selection and pre-submission. Three Notified Bodies with designated scope covering MDR Annex VIII Rule 11 software and active AI review capability were shortlisted. Selection weighted demonstrable software review experience above headline quoted timeline — quoted timelines proved to be weakly predictive, while reviewer familiarity with software lifecycle standards proved strongly predictive of query volume. A formal pre-submission meeting was held before technical file lock.

Sequencing ISO 13485 ahead of technical documentation. The QMS was certified before the technical file was submitted, not in parallel. Notified Bodies audit the QMS as a gate; a technical file submitted against an uncertified QMS generates procedural findings unrelated to product merit.

FDA dossier leverage where structurally valid. The 510(k) dossier was mined for reusable content — software architecture description, verification and validation protocols, cybersecurity documentation, and usability engineering records transferred with modest reformatting. Clinical content did not transfer.

2.5 Health Authority and Notified Body Interaction

Notified Body feedback concentrated on four areas, of which three were anticipated and one was not.

On clinical benefit, the Notified Body's position was that the 510(k) substantial equivalence argument had no standing under MDR. Substantial equivalence to a predicate is a comparative regulatory construct; MDR Article 61 and Annex XIV require demonstration of clinical benefit for the device itself. The reviewer requested direct performance evidence with defined sensitivity and specificity endpoints against an adjudicated reference standard.

On equivalence, the manufacturer's fallback proposal — to claim equivalence to a CE-marked competitor algorithm — was rejected on the grounds established in MDCG 2020-5: equivalence requires demonstration of technical, biological, and clinical equivalence, and the manufacturer must have sufficient access to the comparator's technical documentation. A commercial competitor's algorithm documentation was self-evidently unavailable. This closed the equivalence route definitively.

On PMCF, the reviewer noted the absence of any post-market clinical follow-up framework. Because the US regulatory system has no direct PMCF analogue, the manufacturer had no existing programme to adapt.

The unanticipated finding concerned claims traceability. The reviewer conducted a claim-by-claim mapping of marketing and IFU language against the clinical evidence file and identified several claims — including a quantified workflow efficiency assertion carried over from US commercial materials — that had no corresponding evidence entry. Under MDR Article 7, claims must not mislead as to clinical benefit or performance; unevidenced quantified claims are a straightforward non-conformity.

2.6 Pitfalls and Solutions

Pitfall: the clinical data deficit. The 510(k) evidence base was insufficient for MDR.

Solution. A retrospective clinical performance study was executed using de-identified data from three EU hospital sites, with independent cardiologist adjudication establishing the reference standard. A retrospective design against banked ECG data with prospective adjudication was defensible for a diagnostic software device, avoiding a prospective interventional trial that would have added an estimated 14 to 18 months. The evidence strategy was agreed with the Notified Body at pre-submission — before execution, not after.

Pitfall: no PMCF infrastructure.

Solution. A PMCF plan was built around three data streams: a structured post-deployment performance registry capturing algorithm output against confirmed clinical outcome at participating sites; a systematic annual literature surveillance protocol; and a formalised clinician feedback mechanism with defined analysis triggers. Defining quantitative signal thresholds in advance — rather than committing to "monitor performance" — converted the PMCF plan from an aspiration into an auditable procedure.

Pitfall: QMS framework mismatch. FDA Quality System Regulation practice did not map cleanly onto ISO 13485, particularly regarding vigilance procedures, the PRRC function, and economic operator obligations.

Solution. Rather than remediating existing US procedures, an ISO 13485 QMS was implemented natively and the US procedures mapped into it. Retrofitting EU-specific requirements onto a QSR-shaped system had been attempted first and generated more rework than starting from the ISO structure. A PRRC meeting Article 15 qualification criteria was appointed internally.

Pitfall: EU AI Act exposure discovered mid-project.

Solution. Because the device requires Notified Body conformity assessment under MDR, the embedded AI system falls within the EU AI Act's high-risk classification via Article 6(2) and Annex I. Data governance documentation under AI Act Article 10 was constructed in parallel with the MDR clinical file — specifically, a demographic and clinical characterisation of training, validation, and test datasets, with documented assessment of representativeness against the intended EU patient population. This surfaced a genuine limitation: training data was predominantly US-sourced with limited representation of certain European demographic subgroups. The limitation was disclosed and bounded in the intended purpose statement rather than concealed, and a PMCF objective was added to monitor subgroup performance post-launch.

Transferable lesson. FDA clearance accelerates the engineering and software-validation portions of an MDR submission and does approximately nothing for the clinical and post-market portions. Manufacturers should budget the clinical and PMCF workstreams as though no prior approval existed.

3. Case Study 2 — IVDR Class C Oncology Assay and the Research Use Only Trap

3.1 Client Profile

3.2 Project Scope and Duration

Scope covered intended purpose definition, IVDR classification, ISO 13485 implementation, analytical and clinical performance evaluation strategy, technical documentation, Notified Body engagement, and resolution of the RUO exposure.

Duration: 26 months. The extended timeline reflected a seed-stage starting position with no quality infrastructure and a requirement to generate clinical performance evidence from a standing start.

3.3 Product Characteristics

3.4 The Registration Process and Regulatory Strategy

The IVDR transition fundamentally altered the position of assays of this type. Under the former IVDD, a substantial majority of IVDs were self-certified; under IVDR, the great majority require Notified Body involvement. This assay, informing targeted therapy selection, classified as Class C under Rule 3(f) and therefore required Annex IX conformity assessment with Notified Body performance evaluation review.

The strategic priority was correcting a definitional problem before any technical work began: the manufacturer had not articulated a single, bounded intended purpose. Academic collaborators used the assay across several tumour types with varying gene subsets. Under IVDR, performance must be demonstrated for the intended purpose as stated, and an unbounded intended purpose creates an unbounded evidence obligation.

The intended purpose was narrowed to a defined tumour type set and a fixed gene panel. This was commercially unwelcome and regulatorily essential: it converted an unachievable evidence requirement into a tractable one, with panel and indication extension deferred to post-certification change management.

Performance evaluation was structured per IVDR Annex XIII across the three required components: scientific validity of the analyte-condition association (established through systematic literature review, given well-characterised biomarkers); analytical performance (executed in-house against reference materials and characterised clinical samples); and clinical performance (generated through a retrospective study using banked, annotated FFPE specimens with outcome linkage).

3.5 Health Authority and Notified Body Interaction

Notified Body feedback was dominated by one structural issue that the manufacturer had not recognised as an issue at all.

The Research Use Only designation applied to the interpretation software was challenged directly. The manufacturer's reasoning had been that the sequencing platform was already CE-marked, the software was pre-commercial, and RUO labelling therefore covered the gap. The Notified Body's position was that RUO status under IVDR Article 2(45) and the associated guidance is determined by intended purpose, not by label text or commercial stage. Where software is used in combination with a CE-marked instrument to generate output that informs clinical therapy decisions for identifiable patients, the combination functions as an IVD irrespective of the RUO marking. Labelling a functionally diagnostic component "Research Use Only" does not remove it from IVDR scope; it creates a compliance exposure while providing no protection.

The reviewer further noted that the academic collaborator laboratories were returning variant interpretations into patient care pathways. This was material: it indicated the combination had arguably already been placed on the market in a diagnostic capacity, raising a retrospective compliance question distinct from the forward-looking certification project.

Secondary feedback addressed analytical performance completeness — specifically insufficient characterisation of limit of detection across the variant allele frequency range relevant to FFPE specimens, and inadequate documentation of specimen-handling variability.

3.6 Pitfalls and Solutions

Pitfall: the RUO combination trap. The most consequential finding in the engagement, and one the manufacturer had considered settled.

Solution. A formal intended purpose and combination assessment was conducted, documenting the functional boundary between the CE-marked instrument and the software module and determining that the combination constituted a Class C IVD. Two immediate corrective actions followed: collaborator laboratory arrangements were restructured so that results were not returned into clinical care pathways pending certification, documented via revised written agreements; and RUO labelling was withdrawn and replaced with accurate pre-certification status labelling. The retrospective exposure was documented in a CAPA record with a written rationale — the objective being to demonstrate to the Notified Body a manufacturer who had identified and corrected a problem, rather than one who had concealed it. Voluntary disclosure was assessed as lower-risk than discovery during audit.

Pitfall: the cost-scaling problem for early-stage companies.

Solution. IVDR compliance carries substantially fixed costs — QMS certification, performance evaluation, Notified Body fees — which fall disproportionately on a single-product company. Three measures were applied. First, the QMS was scoped proportionately: ISO 13485 requires a system appropriate to the organisation's activities, not the procedural apparatus of a large manufacturer, and an over-engineered QMS at 12 FTE is itself an audit liability. Second, performance evaluation was sequenced to front-load scientific validity and analytical performance, both lower-cost, establishing whether the clinical performance investment was justified before committing to it. Third, regulatory milestones were mapped explicitly onto the financing plan so that Notified Body queue time was visible to the board as a capital-consumption event.

Pitfall: the retrofit multiple. The company had generated approximately 18 months of development data outside any controlled QMS.

Solution. Pre-QMS data was formally assessed for retrospective qualification. A proportion of analytical work was recoverable through documented retrospective review where raw data, instrument records, and operator documentation were intact; the remainder required repetition under controlled conditions. The recoverable-versus-repeat determination was made jointly with the Notified Body at pre-submission rather than assumed. The general finding — consistent with the practitioner literature's description of a compounding or "snowball" effect — is that early quality investment costs materially less than late remediation, and the differential grows with elapsed development time.

Transferable lesson. RUO is a statement about intended purpose, not a regulatory shelter. Any diagnostic system assembled from a CE-marked instrument plus manufacturer software requires a formal combination assessment before RUO labelling is relied upon.

4. Case Study 3 — Legacy MDD Orthopaedic Implant Transitioning to MDR Class III

4.1 Client Profile

4.2 Project Scope and Duration

Scope covered MDR gap assessment, CER reconstruction, PMCF programme establishment, technical documentation restructuring to MDR Annex II/III, GSPR mapping, and Notified Body design dossier submission.

Duration: 31 months. The longest of the five cases. Legacy transitions are frequently underestimated precisely because the product is commercially mature and the manufacturer reasonably assumes a well-documented history constitutes adequate evidence.

4.3 Product Characteristics

4.4 The Registration Process and Regulatory Strategy

Class III implantable devices attract the most demanding MDR route: Annex IX conformity assessment including design dossier examination, mandatory Notified Body review of the clinical evaluation, and — under Article 54 — potential clinical evaluation consultation procedure involving an expert panel.

The strategic problem was that the manufacturer's clinical evidence architecture was MDD-shaped. The MDD-era CER relied substantially on literature relating to a predecessor design and on the manufacturer's own registry participation, structured as a narrative safety argument. MDR requires a systematically constructed, traceable evaluation with explicit benefit-risk determination and defined clinical benefit endpoints.

The strategy accepted a difficult premise early: the existing CER could not be remediated incrementally and required reconstruction. Attempting to patch an MDD narrative into MDR compliance is a common and expensive error, because MDR non-conformities are structural rather than content-level.

Reconstruction proceeded on three tracks. Legacy clinical and registry data — 15 years of it, and genuinely valuable — was systematically catalogued and assessed for MDR admissibility. A protocol-driven systematic literature review was executed with documented search strategy, pre-specified inclusion and exclusion criteria, screening decisions recorded at article level, and appraisal against defined criteria. A PMCF programme was established prospectively, using national arthroplasty registry participation as its principal data stream.

4.5 Health Authority and Notified Body Interaction

Notified Body feedback on the design dossier was extensive and centred on evidentiary traceability rather than device safety — a distinction the manufacturer initially found difficult to accept, given a 15-year clinical record without significant safety signal.

On literature methodology, the reviewer's central criticism was that the MDD-era literature review presented conclusions without a reproducible audit trail. Search strings were not documented, screening decisions were not recorded at article level, and appraisal criteria were not pre-specified. The reviewer's position, consistent with MEDDEV 2.7/1 Rev. 4 expectations, was that a literature review whose screening cannot be reproduced cannot support a benefit-risk conclusion. This mirrors what the practitioner literature identifies as the single most common CER failure mode: literature summaries presented as literature reviews, with traceability gaps between source data and stated conclusion.

On equivalence, the manufacturer's argument regarding its own predecessor design was partially accepted — an important asymmetry. Because the manufacturer held complete technical documentation for its own predecessor device, the MDCG 2020-5 data access condition was satisfied. However, the reviewer required detailed justification of technical equivalence given a change in polyethylene cross-linking process between generations, and declined to accept clinical equivalence for the subset of claims relating to long-term wear performance, where the material change was directly relevant.

On CER structure, the reviewer characterised the document as insufficiently integrated: PMCF findings, vigilance data, and literature conclusions were presented in parallel sections without demonstrated influence on the benefit-risk determination. The requirement is that the CER function as a living document in which post-market data demonstrably updates the clinical conclusion.

On GSPR mapping, the reviewer required explicit evidence linkage for each applicable Annex I requirement. The MDD Essential Requirements checklist mapped only partially onto Annex I, and several GSPRs — particularly those addressing usability and information supplied to the user — had no corresponding evidence entries.

4.6 Pitfalls and Solutions

Pitfall: literature review traceability failure.

Solution. The review was re-executed under a written protocol specifying databases, search strings with date stamps, PRISMA-consistent screening flow with article-level decision recording, pre-specified appraisal criteria, and an explicit synthesis method linking appraised evidence to each clinical claim. The additional cost was significant. The determinative insight was that the Notified Body was not disputing the conclusion — it was disputing that the conclusion had been demonstrated. Reproducibility, not favourability, was the deficiency.

Pitfall: the living CER problem at scale. The reconstructed CER exceeded 500 pages, and each PMCF cycle, registry data release, and vigilance signal required propagation through it.

Solution. The CER was restructured modularly, separating stable content (device description, state of the art, scientific background) from dynamic content (PMCF findings, vigilance analysis, benefit-risk determination). Update procedures were defined so that a new registry data release triggered revision of specified sections with a controlled propagation path to the benefit-risk conclusion, rather than an unstructured full-document rewrite. A documented change-control log recorded which post-market input drove which conclusion change — directly addressing the reviewer's integration criticism.

Pitfall: no prospective PMCF programme. Fifteen years of retrospective registry participation did not constitute PMCF under Article 61(11) and Annex XIV Part B.

Solution. A prospective PMCF plan was established with defined objectives, pre-specified endpoints including revision rate thresholds benchmarked against national registry averages, defined analysis intervals, and documented escalation triggers. Existing registry relationships were leveraged as the data source, materially reducing cost relative to a de novo post-market study, but the analytical framework was built fresh.

Pitfall: organisational underestimation. The engineering and commercial functions initially treated the MDR transition as a documentation exercise for the regulatory department.

Solution. Design and process input was required from engineering to close GSPR gaps, and claims revision required commercial sign-off. Establishing a cross-functional transition team with explicit executive sponsorship at project initiation would have compressed the timeline; establishing it at month nine, as occurred here, did not.

Transferable lesson. A long, clean commercial safety record is not a substitute for a methodologically traceable clinical evaluation. Legacy manufacturers should assume CER reconstruction rather than remediation, and should budget accordingly.

5. Case Study 4 — German Digital Therapeutic: MDR Class IIa and the DiGA Reimbursement Pathway

5.1 Client Profile

5.2 Project Scope and Duration

Scope covered MDR classification, ISO 13485 implementation, clinical evidence strategy, CE marking, and subsequent BfArM DiGA application under §139e SGB V.

Duration: 24 months total — approximately 14 months to CE certificate, followed by a 10-month DiGA application and provisional listing process. These are sequential, not parallel, and the sequencing is not optional.

5.3 Product Characteristics

5.4 The Registration Process and Regulatory Strategy

This case illustrates a structural feature of the European market that manufacturers consistently underestimate: the CE mark confers the right to place a device on the market, not the right to be paid for it. In Germany these are two distinct processes, before two distinct bodies, with different evidentiary standards.

Classification under Rule 11 was Class IIa. The device provides information used to make therapeutic decisions, but the therapeutic decisions it informs do not carry risk of serious health deterioration in the Rule 11 Class IIb sense, and it does not monitor vital physiological parameters. Class IIa requires Notified Body involvement but permits technical documentation assessment on a sampling basis under Annex IX — a materially lighter burden than the Class IIb full-file review in Case Study 1, and one reason the Rule 11 boundary carries such commercial weight for digital health developers.

The regulatory strategy sequenced three stages. Stage one established the QMS and secured the Class IIa CE mark. Stage two prepared the DiGA application to BfArM under §139e SGB V, which requires the CE mark as an entry condition — a DiGA application cannot precede certification. Stage three targeted the DiGA fast-track provisional listing route, under which a device may be provisionally listed for a twelve-month period during which the required positive healthcare effect evidence is generated, with reimbursement flowing during that window.

The fast-track election was the pivotal commercial decision. Provisional listing permits reimbursed market presence while the definitive trial runs, converting the evidence generation period from a pre-revenue cost into a revenue-supported one.

5.5 Health Authority Interaction

Interaction occurred with two distinct bodies, and the difference in their concerns is instructive.

Notified Body (MDR conformity assessment). Feedback was moderate in volume, consistent with Class IIa sampling. Principal findings concerned software lifecycle documentation under IEC 62304, where the manufacturer's agile development records did not map cleanly onto the standard's expected artefacts; usability engineering under IEC 62366-1, where summative evaluation had not been conducted in the intended use environment with representative users; and clinical evaluation, where the investigator-initiated study was accepted as supporting evidence but required restructuring into a compliant CER with explicit benefit-risk determination.

BfArM (DiGA application). BfArM's concerns were categorically different and substantially more demanding in domains the MDR review had touched only lightly. Data protection and information security assessment was the most intensive element — DiGA requirements under §139e incorporate GDPR compliance, defined information security standards, and specific requirements regarding data processing location and third-party access. Several architectural decisions taken for development convenience, including a non-EU analytics dependency, required re-engineering.

On positive healthcare effects, BfArM required either a demonstrated medical benefit or a demonstrated patient-relevant structural or procedural improvement. The investigator-initiated study was insufficient for definitive listing, which is precisely the circumstance the fast-track route exists to address. BfArM feedback on the proposed trial design was specific and, in the manufacturer's assessment, more prescriptive regarding endpoint selection and comparator choice than the Notified Body's clinical feedback had been.

On interoperability and accessibility, requirements had no MDR equivalent and had not been anticipated in the product roadmap.

5.6 Pitfalls and Solutions

Pitfall: the funding trap. The manufacturer's grant funding had been secured against a project description characterising the product as a wellness and self-management tool — language chosen to fit the funding programme's scope. That description was inconsistent with the medical device intended purpose subsequently required for MDR classification and DiGA eligibility.

Solution. The inconsistency was addressed proactively with the funding body rather than allowed to surface during BfArM review. A formal scope clarification was filed documenting that the underlying technical work was unchanged while the regulatory characterisation had been formalised. This is a recurring and under-discussed hazard for grant-funded digital health developers: language optimised for a funding application can create downstream regulatory inconsistency, and definitional alignment across grant documentation, technical files, and marketing claims should be maintained from the outset.

Pitfall: assuming CE marking equalled German market access. Initial commercial projections assumed revenue commencing at CE certification.

Solution. Financial planning was restructured around the DiGA timeline rather than the MDR timeline, extending the pre-revenue runway assumption by approximately ten months. Germany's health system is additionally federated across more than twenty regional authorities and a plural statutory insurer landscape; DiGA listing provides a national reimbursement route that circumvents much of this complexity, which is precisely what makes it strategically valuable and worth the additional sequential delay.

Pitfall: data architecture incompatible with DiGA requirements.

Solution. Architectural review against §139e requirements was conducted before the DiGA application rather than in response to BfArM findings. The non-EU analytics dependency was replaced and data processing consolidated within the EU. Conducting this review during the MDR phase, when engineering resource was already engaged on compliance work, would have avoided a second re-engineering cycle.

Pitfall: agile development versus IEC 62304 expectations.

Solution. Rather than abandoning agile methodology, the development process was mapped onto IEC 62304 artefacts — sprint documentation structured to generate the required software development plan, requirements traceability, and verification records as process outputs. Agile and IEC 62304 are reconcilable; the failure mode is discovering the mapping requirement after two years of undocumented sprints.

Transferable lesson. In Europe, market authorisation and reimbursement are separate, sequential, and governed by different evidentiary standards. Digital health developers should model the reimbursement pathway timeline from project inception, and should recognise that health technology assessment bodies frequently impose more demanding evidence requirements than Notified Bodies.

6. Case Study 5 — Non-EU Manufacturer, Class IIa Device, and the Forty-Market Problem

6.1 Client Profile

6.2 Project Scope and Duration

Scope covered MDR classification, ISO 13485 gap remediation against an existing MFDS-oriented QMS, authorised representative appointment, technical documentation, Notified Body conformity assessment, EUDAMED registration, and a phased multi-country European market entry plan.

Duration: 22 months to CE certificate and first national market launch. Subsequent national market entries continued on a rolling basis beyond the core engagement.

6.3 Product Characteristics

6.4 The Registration Process and Regulatory Strategy

The classification analysis required care. The device is non-invasive and externally applied, but is used to guide an invasive procedure. Classification follows the device's own characteristics rather than those of the procedure it supports: as an active device intended to allow direct diagnosis or monitoring of physiological processes through imaging, Rule 10 applies, yielding Class IIa. Had the manufacturer's initial assumption of Class I been pursued, self-certification would have been invalid and the resulting market placement unlawful.

The strategic architecture rested on a premise the manufacturer had not held at the outset: the CE mark is a precondition for European market access, not a delivery mechanism for it. Practitioners working across the region describe Europe not as a single market but as forty-plus distinct health systems with separate procurement structures, national registration or notification requirements, language obligations, and reimbursement logic. The CE mark opens the door to all of them and completes access to none.

Four elements were sequenced accordingly.

Authorised representative appointment. Under Article 11, a non-EU manufacturer must designate an EU authorised representative (EC REP). The strategic decision — mirroring the licence-holder question in Asian and Latin American markets — was whether to appoint a commercial distributor or an independent regulatory entity. An independent EC REP was appointed, with distribution arrangements contracted separately. The Article 11 mandate carries legal responsibilities and is named on labelling; where a distributor holds it, changing distributor becomes entangled with regulatory continuity.

EUDAMED registration as a gating item. Actor registration was completed early to obtain the Single Registration Number. Because Actor registration precedes device registration and the non-EU manufacturer workflow requires the authorised representative to be registered and linked first, the SRN sits on the critical path. Following Commission Decision 2025/2371, the Actor, Device, and Market Surveillance modules became mandatory from 28 May 2026; deferring registration pending system maturity was not viable.

Notified Body selection weighted to language and capacity. Selection considered designated scope, realistic capacity, and — materially for this manufacturer — working-language flexibility during assessment.

Phased national market entry. Rather than a simultaneous pan-European launch, three initial national markets were selected on procurement accessibility and reimbursement clarity, with subsequent markets added on a rolling schedule.

6.5 Health Authority and Notified Body Interaction

Notified Body feedback was moderate, reflecting Class IIa sampling assessment, and concentrated on three areas.

On electrical safety and electromagnetic compatibility, IEC 60601-1 and 60601-1-2 documentation was largely acceptable, having been generated for MFDS and FDA submissions, though additional evidence was required for the specific IEC 60601-2-37 requirements applicable to ultrasonic diagnostic equipment.

On usability, the reviewer required summative usability evaluation conducted with representative EU clinical users. Evaluations performed with Korean clinical users were considered insufficient to establish usability in the intended European use environment, on the grounds that clinical workflow and training background differ materially. This finding is frequently unanticipated by non-EU manufacturers.

On clinical evaluation, the existing evidence base — supporting MFDS and FDA submissions — was accepted in substantial part, with additional literature analysis required addressing paediatric use, where the manufacturer's own data was thinner.

Beyond the Notified Body, interaction with national competent authorities proved more administratively demanding than anticipated. Several member states operate national registration or notification requirements additional to EUDAMED. Requirements are not harmonised, procedures are documented in national languages, and processing times vary substantially. This work is unglamorous, sits outside the conformity assessment process, and is routinely omitted from market entry planning.

6.6 Pitfalls and Solutions

Pitfall: treating Europe as one market.

Solution. Market entry was restructured as a phased sequence with country-level requirement mapping — national registration or notification obligations, language requirements, reimbursement route, and procurement structure documented per target market before commercial commitment. Practitioners advising in this area frame the underlying calculation sharply: a six-month delay in market entry can cost roughly a third of five-year cumulative profit on a product line, because delay compresses the revenue-generating portion of the commercial lifecycle while fixed development costs remain unchanged. Under that arithmetic, sequencing decisions are financial decisions, not administrative ones.

Pitfall: language and translation scope underestimated. Member states may require Instructions for Use and labelling in their official language or languages, and full pan-EU coverage can require in excess of twenty language versions.

Solution. Translation was scoped to the phased market plan rather than executed comprehensively upfront, which avoided translating into languages for markets not yet targeted. Critically, translation was managed under document control with a defined revision-propagation procedure — an IFU change must reach every language version, and an uncontrolled translation set becomes an audit finding and a vigilance liability.

Pitfall: usability evidence not transferable across regions.

Solution. Summative usability evaluation was re-executed with representative EU clinical users across the three initial target markets, covering both adult and paediatric probe configurations. The general principle — that human factors evidence is environment-specific and does not transfer across clinical cultures — should be assumed by any non-EU manufacturer entering the EU.

Pitfall: QMS structured for MFDS expectations.

Solution. Gap assessment identified EU-specific requirements absent from the existing system: PRRC designation under Article 15, MDR vigilance procedures with EU reporting timelines, economic operator obligations under Articles 13 and 14, and PMS procedures generating Article 86 PSURs. These were integrated into the existing ISO 13485 system rather than maintained as a parallel EU-specific structure. Parallel QMS structures were considered and rejected: dual systems diverge, and divergence surfaces during audit.

Transferable lesson. For non-EU manufacturers, the CE mark is the midpoint of the European market access project rather than its conclusion. Authorised representative selection, EUDAMED sequencing, national registration mapping, and language scope determine whether certification converts into revenue.

7. Cross-Case Analysis

7.1 Recurring Failure Modes

Across all five composites, four failure modes recur independent of device type, risk class, and company scale.

Clinical evidence is systematically underestimated. In every case, the clinical or performance evidence workstream consumed more time and cost than planned. The mechanism differs — FDA equivalence not transferring, an unbounded IVD intended purpose, an untraceable legacy literature review, insufficient evidence for reimbursement listing — but the pattern holds. MDR and IVDR require demonstrated clinical benefit or performance for the device as specified, and no adjacent approval or commercial history substitutes for it.

Quality system sequencing is misjudged. Manufacturers who certified the QMS before submitting technical documentation encountered materially fewer procedural findings. Those who ran the two in parallel, or who retrofitted an ISO 13485 structure onto a differently-shaped system, generated rework.

Regulatory characterisation drifts across documents. Case Study 2's RUO designation and Case Study 4's grant language are the same failure in different guises: a product described one way in one document and another way elsewhere. Notified Bodies and health technology assessment bodies read across documents, and inconsistency is a finding.

Market authorisation is confused with market access. Cases 4 and 5 illustrate the two principal forms — reimbursement as a separate process with a separate evidentiary standard, and national market fragmentation beneath the CE mark.

7.2 Comparative Summary

7.3 What Prior Approvals Actually Transfer

The five cases permit a reasonably precise statement of what a non-EU approval contributes to an EU submission.

Transfers with modest reformatting: software architecture and lifecycle documentation, electrical safety and EMC test reports against harmonised IEC standards, biocompatibility data against ISO 10993, sterilisation validation, packaging and shelf-life data, cybersecurity documentation.

Transfers partially: risk management files, which require restructuring to ISO 14971 with explicit GSPR linkage; and quality system documentation, where structural mapping is required.

Does not transfer: clinical evidence premised on substantial equivalence; usability evidence generated outside the intended EU use environment; post-market surveillance frameworks where no PMCF analogue exists in the origin jurisdiction; and marketing claims, which require claim-by-claim evidence substantiation under Article 7.

The practical planning implication is that prior approval compresses the engineering and testing portions of an EU submission substantially, and the clinical and post-market portions barely at all.

8. Conclusion

The cases examined here are composites, but the failure modes are not invented. They are the patterns that European regulatory practitioners describe encountering repeatedly across the 2026 market: transatlantic manufacturers discovering that substantial equivalence has no MDR standing; early-stage companies discovering that RUO labelling is a statement of intended purpose rather than a shelter; legacy manufacturers discovering that fifteen clean years do not constitute a traceable clinical evaluation; digital health developers discovering that a CE mark and a reimbursement decision are different achievements; and non-EU manufacturers discovering that Europe is forty markets wearing one mark.

What distinguishes the engagements that resolve within planned timelines from those that overrun is consistently not technical sophistication. It is the timing of three decisions: when the classification determination is made and stress-tested, when the quality system is built, and when the Notified Body is first engaged. Manufacturers who resolve all three at project initiation encounter the EU framework as demanding but navigable. Those who defer them encounter it as a marathon run in the wrong direction, and the remediation cost of a late correction consistently exceeds the cost of early rigour by a substantial multiple.

The framework is not becoming less demanding. The proposed MDR and IVDR reform offers relief in specific areas — Breakthrough Device pathways, removal of fixed recertification cycles, PRRC flexibility for SMEs, predetermined change control plans — but leaves the lifecycle evidence philosophy intact. Planning for that philosophy, rather than hoping to be excused from it, remains the operative professional discipline.

Disclaimer

The case studies in this article are illustrative composites constructed for educational purposes. They do not describe specific client engagements, manufacturers, products, Notified Bodies, or regulatory decisions, and should not be relied upon as precedent. Nothing in this article constitutes legal, regulatory, or professional advice. EU MDR, IVDR, EU AI Act, and national reimbursement requirements are subject to ongoing guidance publication and legislative amendment. Manufacturers should consult qualified regulatory affairs professionals, their designated Notified Body, and the relevant competent authorities for device-specific guidance.