Contents
1. Start with the claim | 6. What belongs in the plan |
2. Clinical evidence sources | 7. Evaluation report structure |
3. Avoiding a new study | 8. Examples by device type |
4. Legacy data | 9. Common review problems |
5. General vs specific methods | 10. FAQ |
Under the European In Vitro Diagnostic Regulation (IVDR 2017/746), a post market performance follow up IVDR strategy requires continuous, proactive data collection such as clinical studies or systematic registration audits rather than passive vigilance monitoring. To survive Notified Body review, this data must be compiled into a highly structured evaluation report that maps real world performance metrics directly against pre defined, quantitative acceptance. Which clinical performance sources hold up, how to use legacy data and literature, when you can skip a new study, and how to write the evaluation report so the logic is easy to follow.
For post market performance follow up IVDR, A lot of PMPF writing stops at Article 56(6) and a list of Annex XIII headings. That does not help much once you are already running PMS and preparing PERs. The harder question is simpler:
WHAT REVIEWERS ARE REALLY CHECKING
For the claims in your IFU, and the gaps still open in your PER, what post-market data is enough and what is only paperwork?
Below is a practical way to decide which sources count, how far legacy data and literature can take you, how to reduce PMPF scope without looking like you are dodging the requirement, and how to organise the evaluation report. The report layout draws on patterns seen in mature Class C serology files (ELISA-type assays with long market history), written in general terms only.
Start With The Claim, Not The Template For Post Market Performance Follow Up IVDR
Article 56 treats performance evaluation as ongoing work, not a one-off pre-market exercise. Article 56(6) requires updates from the PMPF plan (Annex XIII Part B) and the PMS plan (Article 79). For Class C and D devices, the PER must be updated at least once a year. This supports post market surveillance medical device traceability.
IN PRACTICE
The plan only works if it answers specific performance questions. Generic wording about “confirming safety and performance” does not get you far in review.
Problems that usually attract findings
IFU claims wider than the evidence behind them
Open PER or risk-file gaps that never become post-market questions
PMPF limited to complaints and a literature search, with no link to sensitivity, specificity, cut-offs, or scientific validity
A waiver based only on “well-established technology,” with no device-level proof
Literature used as if it studied your kit when it only covers the analyte or another manufacturer’s assay
Build a short gap map first
Put these four inputs next to each other before you draft the plan:
Intended purpose → PER conclusions → Risk-file gaps → SOTA claims → PMPF questions
1. Intended purpose: what is measured, function, clinical context, population, specimen types, intended user, and exclusions (for example, not for donor screening).
2. PER conclusions: scientific validity, analytical performance, clinical performance.
3. Risk management file: residual risks and open questions, including known limits such as IgM persistence, reactivation, or grey-zone results where relevant.
4. State of the art: guidelines, common specifications if any, competitor performance, EQA expectations.
Claim / uncertainty | Pre-market evidence | Gap after CE marking | PMPF question | Lowest-burden method |
Clinical sensitivity ≥98% in symptomatic adults | Single-site residual-sample study, n=180 | No multi-site confirmation; paediatric claim not supported | Does sensitivity hold across EU labs and age bands? | Multi-lab leftover-sample concordance + EQA panels |
Detects current circulating variants | In silico inclusivity + limited wet panel | New lineages keep appearing | Do emerging variants stay within claimed inclusivity? | Quarterly literature/variant checks + targeted wet testing when risk appears |
No biotin interference at OTC doses | EP07 spike study | High-dose biotin use is more common in practice | Are interference complaints or discordants rising? | Complaint coding + targeted survey of high-risk labs |
Qualitative serology still within SOTA | Legacy clinical package + IFU claims | No papers naming the device this cycle | Do guidelines and peer assays still support the intended use and cut-off logic? | Time-bounded literature review + equivalence table + peer-assay vigilance |
SIMPLE RULE
No gap means no need to invent a study. A real gap should not be buried under “general PMS.” That table is what the plan should be built around.
Clinical Evidence Sources And How They Usually Land In Review For In Vitro Diagnostic Medical Devices
Under EU In vitro diagnostic medical devices 2017/746 Article 56(3), clinical evidence is the combination of scientific validity, analytical performance, and clinical performance. Article 56(1) matters more than people use it: you must set and justify the level of clinical evidence for the device and its intended purpose. That is the legal basis for not treating every Class B chemistry assay like a Class D programme if the justification is written properly.
1. Device-specific clinical performance studies: strongest for novel claims, companion diagnostics, Class D, and any claim that cannot be tied to your device through other data.
2. Leftover / residual sample studies: often the most efficient post-market option for IVDs, because you are checking the quality of the result, not intervening on a patient.
3. Routine diagnostic testing data: LIS extracts, discordant-result logs, peer method comparisons. Useful when representativeness and analysis methods are clear.
4. EQA / ring trials / proficiency testing: strong for real-world reproducibility across sites; one of the specific methods Annex XIII Part B points to for IVDs.
5. Published literature and guidelines: usually enough for scientific validity. For clinical performance, only when you can show the data applies to your device: same analyte, similar technology, comparable intended use, and performance metrics you can transfer.
6. User surveys and support inquiries: helpful for usability, off-label use, and rare matrix issues. Weak as the only support for sensitivity or specificity unless the survey is designed like a study.
IF THE IFU STATES A NUMBER
Sensitivity, specificity, LoD, agreement reviewers expect a data path that can still defend that number after market release. Literature alone rarely does that for Class C or D unless it is about your device or a carefully justified equivalent.
When the literature does not name your kit
For older serology assays, a review window often returns no papers on the manufacturer’s own product. That is common. The problem starts when competitor or analyte-level papers are written up as if they were device-specific clinical performance.
When A New Clinical Performance Study May Not Be Needed For PMPF IVDR
Annex XIII Part A, 1.2.3
Clinical performance studies are required unless you give a proper justification for using other sources of clinical performance data. A justification that usually holds includes:
scientific validity is not in dispute
analytical performance is complete and device-specific
clinical performance is supported by literature and/or routine diagnostic experience that applies to this device or a justified equivalent
the intended purpose is not novel (no new population, specimen type, or clinical decision claim)
residual risks do not depend on unproven clinical performance parameters
USUALLY ACCEPTABLE
Class B serum creatinine on a mature platform, years of EQA, published method comparisons for the same reagent system, no open performance trends. PMPF can rest on EQA, literature, and complaint or discordant analysis.
USUALLY REJECTED
Class C molecular assay claiming saliva when the only clinical literature is for NPS or serum. The specimen-type claim creates a gap that literature on another matrix cannot close.
Article 2(39) and devices where clinical performance is not meaningful
For in vitro diagnostic medical devices, MDCG 2022-2 notes that clinical performance data may not be expected for some products, such as non-sterile specimen receptacles, microscopy slides, or certain general reagents. You still need a performance evaluation and a written justification. Reduced PMPF should focus on functional or analytical performance and safety signals, not invented diagnostic accuracy endpoints.
Annex XIII Part B, Section 8 and the more common outcome
The regulation allows a documented case that PMPF is not appropriate for a specific device. Full waivers are uncommon. What mature Class C files more often conclude is:
TYPICAL WORDING THAT CAN WORK
General PMPF methods remain in place. No additional dedicated PMPF study is needed at this time.
That only works if the report shows the supporting package, not just the sentence:
Evidence element | Role in the argument |
Time-bounded literature and guideline review | Scientific validity and SOTA are still current |
Equivalence / similar-device table | Claims sit within the peer performance range |
Vigilance on peer assays (MAUDE, HPRA, FSCA) | Class risks are watched even if your kit has no events |
Complaints and support cases coded by performance impact | Diagnostic failures are separated from packaging issues |
Exposure denominator (kits sold or estimated tests) | Complaint counts can be interpreted |
CAPA / internal reports with performance impact | Manufacturing issues closed without residual performance risk |
Stability through labelled expiry | Analytical performance holds for the claimed lifetime |
No FSCA or serious incidents for the device | Benefit–risk is not challenged by vigilance |
Intended-use boundaries monitored | Off-label or misuse signals are tracked |
RESTART TRIGGERS
New intended purpose, design or cut-off change, rising performance complaints, guideline shift, peer-class FSCA pattern, EQA failure trend. Without triggers, “no additional study” looks like a permanent opt-out.
Articles 57, 58 and 70
If you do need new post-market data, study design drives the regulatory load (see MDCG 2025-5):
Situation | IVDR frame | What it means day to day |
Any performance study | Article 57 | Ethics and scientific integrity still apply |
Non-CE or off-label use with invasive sampling, interventional use, or added risk | Article 58 | Full authorisation pathway |
CE-marked within intended purpose, no extra invasive or burdensome procedures | Art. 57; often no Art. 70(1) | Leftover-sample PMPF often sits here |
CE-marked within purpose with extra invasive or burdensome procedures | Article 70(1) | Notification required |
Companion diagnostics using leftover samples | Art. 58(2) carve-out | Evidence expectations remain high |
One practical way to limit burden without skipping evidence is to design PMPF studies on leftover samples within the intended purpose, with pre-set endpoints, instead of defaulting to interventional designs that pull you into Article 58 or 70. This keeps post market performance follow up IVDR evidence focused and allows PMPF IVDR studies to remain proportionate.
Legacy Data Under MDCG 2022-8 For Post Market Surveillance IVDR
Long market history helps the proportionality argument. It does not replace the PMPF evaluation report. Treat IVDD-era material as input to appraise, not as finished IVDR evidence.
Legacy asset | How to use it | Common review pushback |
Old clinical performance studies | Map to Annex XIII characteristics; check quality, bias, population, comparator | Not representative of current purpose or EU population |
Analytical verification | Re-present against current GSPRs and claimed specimen types | Missing modern interferents or incomplete precision design |
Literature files | Re-run with search strategy, inclusion rules, appraisal, and applicability | Unstructured collection of supportive papers |
Complaints and vigilance | Turn into performance trends and PMPF triggers; separate packaging from diagnostic issues | No link to performance claims |
EQA participation | Use as an ongoing method with acceptance criteria | Certificates filed but never analysed |
Stability programmes | Link labelled lifetime to lot data through expiry | Lifetime claimed but not shown post-market |
For post market surveillance IVDR, MDCG 2022-2 is clear that IVDD clinical performance studies are “other sources of clinical performance data.” They do not automatically meet Annex XIII 2.3. Quality and completeness still need assessment, usually with literature and/or routine diagnostic experience alongside them.
1
List every performance claim in the current IFU, including grey-zone rules and exclusions.
2
Rate each claim as strong, partial, or weak against the legacy package.
3
Close gaps that cannot wait for post-market work (novel claims, Class D common specifications, CDx bridging, new matrices).
4
Move acceptable residual uncertainties into PMPF questions with timelines.
5
Use the first PMPF evaluation report as the base for the future IVDR PER.
General Methods Vs Specific Methods For Post Market Surveillance Medical Device Evidence
Annex XIII Part B expects methods for proactive collection of safety, performance, and scientific data. In the absence of an IVD-specific template, many teams adopt MDCG 2020-7 and 2020-8 from PMCF. These methods support post market surveillance medical device evidence and keep post market surveillance IVDR planning connected to the PER.
Overlap with PMS and PSUR is normal. What matters is traceability: which outputs update the PER, risk file, IFU, and SSP for Class C and D, plus a clear statement on impact to technical documentation. In post market surveillance medical device documentation, post market surveillance IVDR traceability should remain clear.
Mature Class C qualitative serology: often general methods plus EQA and stability, with a justified “no additional study” conclusion if performance complaints are absent and SOTA is stable. Peer-assay problems in public databases should still be discussed as class risk.
Infectious disease molecular assays: usually need variant or strain surveillance and wet confirmation when in silico risk appears.
Quantitative clinical chemistry: EQA against peer methods often beats a weak survey.
Companion diagnostics: concordance with the clinical trial assay logic, label changes, and new indications.
What Belongs In The PMPF Plan For Post Market Performance Follow Up IVDR
For a mature qualitative serology kit, the question can be softer but still measurable: no performance-related complaint trend above a set rate per kits sold; EQA outcomes within IFU interpretation rules; no guideline change that makes IgM-only use without reflex testing unacceptable.
Each method linked to a question, owner, analysis approach, and decision rule
Direct references to PER and risk management sections
Schedule aligned with Article 56(6) and PSUR timing (annual for Class C and D)
A real similar-device table under Annex XIII Part B 5.2(f): principle, analyte, cut-off or grey zone, specimen, procedure differences, published Se/Sp, comparability score
Clear reasons for methods you are not running
Evaluation Report Structure For IVDR Performance Evaluation
The report should read as a decision record for the cycle, not a reprint of the PER and not a dump of raw tables.
Scope → Activities → Literature (two streams) → GSPR mapping → Complaints → CAPA / stability / FSCA → TD impact → Conclusion
Literature: keep two jobs separate
Cycle PMPF literature: short window (for example 6–12 months): safety signals, peer-assay problems, recalls.
Scientific validity refresh: analyte condition association, guidelines, longer look-back if needed (Annex XIII Part A 1.2.1).
For each stream, record databases, dates, keywords, inclusion and exclusion rules, numbers retrieved and included, appraisal, and whether any hit is device-specific.
Map findings to GSPRs
GSPR 1 intended performance and safety under normal use
GSPR 5 use-error risks (trained user, IFU clarity, grey-zone and retest rules)
GSPR 8 benefit–risk still acceptable; known limits managed (for example IgM not used alone)
Complaint coding
Bucket | Examples | What it means for PMPF |
Diagnostic performance | False reactive or non-reactive results, invalid rates, cut-off disputes | Direct challenge to claims |
Reagent / kit physical quality | Leakage, particulates in controls, shipping damage | Manufacturing or CAPA issue; check whether performance was affected |
Use / instrument setup | Invalid runs linked to washer or reader setup | May point to IFU or training actions under GSPR 5 |
Documentation only | No product defect | Record for completeness |
ALWAYS INCLUDE
An exposure denominator (kits sold or estimated tests) for the same period, and a cross-reference to the PSUR for Class C and D so the two documents do not diverge.
Close with an explicit technical documentation impact line: no change to IFU, PER, risk file, or design or named updates with owners. A conclusion that the device “remains safe and effective” without that line leaves the Article 56(6) loop incomplete.
Examples By Device Type
Common Review Problems
What gets rejected | Why | What to do instead |
PMPF limited to complaints | Does not confirm performance proactively | Add EQA, concordance, SOTA or equivalence work, or targeted studies |
One generic plan for the portfolio | No device-specific questions | Build the gap map from PER, risk file, and IFU |
Literature without applicability | Cannot support your numbers | Separate scientific validity, SOTA, and device-specific data |
Complaint counts without exposure | Rate cannot be interpreted | Report kits sold or tests in the same window |
Packaging issues counted as clinical performance | Confuses the performance story | Code by diagnostic impact |
“No additional study” with no supporting package | Looks like avoidance | Show the bundle, triggers, and next review date |
No technical documentation impact line | Breaks the Article 56(6) loop | State change or no change, with owners if needed |
PMPF and PSUR tell different stories | Traceability failure | Share source tables and cross-references |
10. Working Approach
1
Claims drive evidence obligations. Narrow claims and clear exclusions reduce work more than a clever waiver.
2
Gaps create questions; questions create methods. Methods without questions are filler.
3
Legacy data is usable after appraisal. Market age supports proportionality; it does not skip the report.
4
Interventional studies can often be avoided. The performance question cannot.
5
Mature Class C serology can support “no additional study” only with a full general-methods package and restart triggers.
6
Article 56(1), Annex XIII 1.2.3, and Part B.8 work when written as evidence arguments.
7
The evaluation report should show: activities, evidence, GSPR and benefit–risk impact, technical documentation impact, and next actions.
IF YOU ONLY DO ONE THING
Open the PER, list every claim that is only partly supported, and assign each claim a post-market method, an owner, and a decision rule. That list is the PMPF plan.
FAQ
What is PMPF under IVDR?
Can PMPF be waived?
When can I avoid a new clinical performance study?
Does PMPF apply to legacy IVDs?
What evidence sources do Notified Bodies usually prefer?
Which articles apply to leftover-sample PMPF studies?