How not to leave PMS to the last minute as a medical device manufacturer
These articles reflect our opinions and thoughts on the subjects covered, not purely factual statements. We encourage readers to consult multiple sources and their own medical professionals.
If you sell a medical device in Europe and you are trying to keep a CE mark without burning the budget on a last-minute data scramble, this is for you. The short version: do not leave post-market surveillance (PMS) to the last minute, and you avoid a lot of the wrath of the MDR.
The Medical Device Regulation (MDR - EU 2017/745) is already live. Deadlines for custom-made implantable devices have passed. Higher risk classes keep following. As products migrate from the old Medical Device Directive (MDD) to MDR, one expensive stumbling block keeps showing up: post-market surveillance.
Under MDD you could lean harder on adverse events that arrived of their own accord, and on performance work done before launch. Under MDR the word on the street is proactive.
Plain English
- Post-market surveillance (PMS)
- The ongoing job of watching how your device behaves once it is sold.
- Post-market clinical follow-up (PMCF)
- The clinical slice of that work: keep collecting evidence on safety and performance after launch.
- Real-world evidence (RWE)
- Outcomes from patients in ordinary clinic, not a carefully controlled trial.
- Product-linked registry
- Follow-up where the questionnaire and any adverse event sit next to a named SKU, not just “knee replacement” as a class.
What proactive PMS actually means under MDR
Proactive post-market surveillance means you keep looking for safety and performance issues in products that are already on the market. You do not wait for the complaint form to land.
There are understandable reasons a company would rather not. You spent a fortune developing and launching the thing. Digging for reasons to take it off the market feels backwards.
That is roughly what the regulation asks for. Keep looking. If you do not find a problem, you keep selling. If you never look properly, the notified body eventually notices.
MDD-era habit
Mostly reactive
Wait for adverse events. Lean on pre-launch performance data. Hope the file holds.
MDR expectation
Proactive follow-up
Keep collecting safety and performance data in normal use, tied to the product you sell.
As I write this in September 2026, I keep hearing the same scramble: companies short on PMS data throwing money at regulatory consultants and CROs, hurrying ethics boards, and hiring researchers to gather data by hand so a notified body will stay quiet. It is easy to avoid that mess. It is also disastrously attractive to kick the can down the road.
There is a better route than a last-minute manual harvest. Real-world evidence, gathered early, on a schedule patients will actually answer.
If you want the commercial argument for why that same file can also help market access, I already wrote the data you need to sell is the data MDR already wants. This piece is about the registry-shaped gap underneath it.
Safe and effective does not have to mean a seven-figure trial
As a manufacturer you still need to show two things: the product is reasonably safe, and it does what it claims in the people who actually get it. That can be a classic randomised trial that runs into seven figures. It does not have to be.
MDCG 2020-7 is the PMCF plan template. Registries and planned real-world analyses show up there as legitimate methods, next to studies and surveys. A thin satisfaction survey is usually not what gets you out of trouble on a higher-risk device.
Why registries suit post-market follow-up
When people say “registry”, they often picture an externally funded national database used to gauge treatment performance. Those exist in some specialties. The UK National Joint Registry (NJR) for hip and knee replacement is the famous example.
Done properly, registries have three advantages that make them a natural fit for MDR-style PMS.
Care teams enrol patients
Invite, demographics, contact details. Patients fill in the questionnaires. Far less paid researcher time copying notes.
The data is longitudinal
Email and phone mean reminders. You can show change over months, and give patients a route to report problems.
Lighter governance than a trial
Often treated as routine care, audit, or service evaluation rather than a full research ethics route. Still check local rules.
Care teams enrol patients
In a classic clinical study you often pay researchers to copy notes and chase questionnaires. In a modern registry-style setup the care team invites the patient, the patient adds a bit of demographic and contact information, and then they get reminded to complete questionnaires on a schedule.
That is way less work for researchers, often none at all on the chase, and routinely tens of thousands of pounds off the manufacturer bill compared with standing up a traditional study infrastructure for every SKU.
The data is longitudinal
Because the patient gave a phone number or email, you can remind them. That is how you show, for example, that a new hip implant looks equivalent or better than the current state of the art over time. It is also how you give patients a chance to report adverse events so you can argue safety with something stronger than silence.
The hard part is not picking the form. It is getting the second and third completions back. We wrote about that in 7 Proven Strategies to Boost Patient Engagement with PROMs.
Lighter governance than a trial - with a caveat
When a care team gathers outcomes as part of routine care or to audit their own results, you can often skip the months of delay and the five-figure ethics consultancy bill that come with a formal research study. That is one of the reasons registries are attractive for PMS.
It is not a free pass. Rules differ by country and by how you frame the work. If you are publishing research, sharing identifiable data outside the care team, or running something that looks like an interventional study, you may still need ethics or equivalent approval. Check before you assume.
The catch with existing registries
Existing registries do not exist for most procedures. Where they do exist, even fewer are usefully product-linked for your commercial portfolio.
Where the big European registries sit
They are densest where high-volume, high-cost implants already had political will: joint replacement, mostly. That is brilliant if you make those implants. It is not a PMS plan for an injectable, a soft-tissue device, a consumable, or half the Class IIa and IIb catalogue.
| Registry | Region | Focus | Manufacturer caveat |
|---|---|---|---|
| NJR | UK | Hip, knee, ankle, elbow, shoulder arthroplasty | Strong implant tracking. Not your HA syringe or most non-arthroplasty devices. |
| EPRD | Germany | Hip and knee arthroplasty | Very granular product library. Still joint-heavy. |
| LROI | Netherlands | Arthroplasty | Product numbers and barcode workflows. Same specialty concentration. |
| RENACOT | France | National orthopaedic registry (SOFCOT) | Hip, knee, shoulder and expanding. Still not a general medtech PMS net. |
Snapshot for orientation, not a complete European census. Coverage and implant libraries change. Verify against the registry you intend to rely on.
Product linking is the point
MDR, quite rightly, cares which product was used, and how that relates to outcomes and adverse events. Knowing that most total knee replacements do well does not prove your knee implant is safe.
Not enough
“Knee replacements have good outcomes”
Procedure-level. Could be anyone’s implant. Hard to put in your PMCF file as proof for your SKU.
Product-linked
“SKU X, indication Y, Oxford at 12 months”
Named device, population, timepoint, and a real questionnaire. That is a sentence a notified body can read.
What to do if no product-linked registry exists
Just make your own.
Disclaimer: what follows is how we at Patient Watch create registries for clients. It is our experience gathering real-world data, why it is cheaper than traditional registry builds, and why that shape fits post-market surveillance. It is not a regulatory opinion, and it is not a substitute for your notified-body dialogue.
Traditionally, a national-scale registry budget sits in seven figures. One public benchmark: the UK NJR data-management and statistics tender sits in the tens of millions over a multi-year contract (core figures published around the mid-teens of millions including VAT for five years, before extensions). That is clearly more than a manufacturer can spend per SKU. HQIP’s NJR tender page is the trail if you want the numbers.
Those programmes were expensive partly because the software was bespoke. Paying programmers to build a regulated, use-case-specific system from scratch is a long and costly job. Most manufacturers do not need a national implant library. They need a product-linked follow-up for the devices they actually sell.
What we do differently
Patient Watch was designed for clinicians to track recovery along a pathway: does the patient get better after the doctor does something. I built it hoping one platform could serve my mum (a GP) and my dad (an orthopaedic surgeon), which meant the data model had to flex across specialties.
The parameters vary, but the shape is usually the same three blocks the MDR already cares about:
1. Demographics and symptoms
Age, sex, BMI, primary symptoms, symptom duration - the population you will later filter on.
2. Intervention and product
The surgery, injection, or device used - including the SKU - so outcomes are not orphaned from the product.
3. Patient response over time
Questionnaires on a schedule, adverse-event prompts, quality of life - safety and performance in the patient’s own words.
That is already a registry-shaped dataset. Since launch we have added study grouping across clinics and countries, each study tied to specific product SKUs, with exports and charts that care teams and quality people can both read. Behind that sits the boring but necessary stuff: ALCOA+ style integrity expectations, GDPR, role and organisation access, and pseudonymisation so manufacturer sponsors do not get patient names with the outcomes.
For a concrete follow-up shape in knees, see the knee registry follow-up model. For questionnaire choice when you are publishing or benchmarking, see which knee questionnaire to use.
Kick the can, or start the schedule
Wait
- Notified body asks for PMS you do not have
- Consultants and CROs on rush rates
- Ethics and manual capture under pressure
- Data that still may not be product-linked
Start now
- Invite patients as they are treated
- Questionnaires on a schedule with reminders
- SKU sitting next to every outcome
- A file that grows while you keep selling
More on how we run the capture side: post-market surveillance and real-world evidence.
Our take
Most MDR post-market data gaps are registry gaps. If a national, product-linked registry already covers your implant, use it. If it does not - and for most devices it will not - build a light, product-linked follow-up before someone else bills you to invent one in a hurry.
FAQ
Is a national registry enough for MDR?
Sometimes, if it covers your procedure and links outcomes to your product in a way your notified body accepts. For most SKUs outside high-volume arthroplasty, the national registry either does not exist or does not speak your product’s name. Then you still need your own product-linked follow-up.
Do I need a full RCT for PMCF?
Not always. MDCG 2020-7 lists registries and planned real-world analyses among the methods. What you need depends on the device, the gaps in your clinical evaluation, and what your notified body will accept. I am not signing that off for you.
When is ethics still needed?
When the work is framed as research, when sharing goes beyond what routine care or audit allows, or when local rules say so. “Registry” is not a magic word that deletes governance. Ask someone who knows your country before you promise “no ethics”.
What does product-linked mean in practice?
The outcome row knows the SKU (or equivalent product identifier), the indication, and the timepoint. “Knee injection got better” is not product-linked. “Our HA, in this age band, at six months on WOMAC” is.
Do you replace our regulatory writer?
No. We help gather the questionnaires and structure the study. Your regulatory people still write the dossier.
Sources
- Regulation (EU) 2017/745 on medical devices (MDR).EUR-Lex
- MDCG 2020-7. Post-market clinical follow-up (PMCF) Plan Template. April 2020.PDF
- National Joint Registry.njrcentre.org.uk
- HQIP. NJR data management tender information (2026).hqip.org.uk
- EPRD (German Arthroplasty Registry).eprd.de
- LROI (Dutch Arthroplasty Register).lroi.nl
- SOFCOT / RENACOT.sofcot.fr
Guy built Patient Watch at Imperial College London for his father, an orthopaedic surgeon. Today, NHS and private clinics use Patient Watch to automate questionnaire schedules, track patient recovery after injections and surgery, and collect real-world clinical evidence.
Related articles
The data you need to sell is the data MDR already wants
How product-linked follow-up can support both post-market clinical follow-up and market-access conversations.
A knee registry that tracks recovery hard, then checks in lightly to year five
Denser PROMs in year one, light annual follow-up to year five, shared across arthroplasty, sports, and patellofemoral care.
Decentralised clinical studies: collecting real-world data without the CRO price tag
A practical guide for medical device manufacturers and clinical researchers running remote longitudinal studies and PMCF data collection.