As per MDR Article 83, “for each device, manufacturers shall plan, establish, document, implement, maintain and update a post-market surveillance system in a manner that is proportionate to the risk class and appropriate for the type of device...

The post-market surveillance system shall be suited to actively and systematically gathering, recording and analysing relevant data on the quality, performance and safety of a device throughout its entire lifetime, and to drawing the necessary conclusions and to determining, implementing and monitoring any preventive and corrective actions...”.

Post-market clinical follow-up (PMCF) can be done in many different ways, whereby clinical studies are one. The post-market data is relevant in that it feeds into the continued work of keeping the device’s benefit-risk profile up to date, making sure that the CE mark stays valid. The data collected will go into a wide range of reports (partly depending on device class), including:

• The Risk Management Report
• The Post Market Surveillance Report
• The Periodic Safety Update Report
• The Summar of Safety and Clinical Performance Report
• The Clinical Evaluation Report.

In case you do need to perform a clinical study, it is necessary to decide what kind of study you want to do. ISO14155:2026, the GCP Standard for medical devices, describes in Annex I the clinical development stages in the following way:

The post-market stage can include clinical investigations to gather additional confirmatory data, but can also include observational clinical investigations, with the additional benefit of unveiling information on rare adverse events and long-term outcomes.

The post-market clinical investigation may include both retrospective as well as prospective data collection. Whether retrospective or prospective, it is necessary to establish which data you intend to collect.

If you now decide to collect Standard of Care (SoC) data, there are several known pitfalls that need to be addressed.

When collecting Standard of Care (SoC) data in PMCF studies, it is done to generate contextual, real-world evidence to support the ongoing assessment of safety, clinical performance, and benefit–risk under MDR.

Although the SoC data may be weaker when it comes to comparisons (and you should be transparent about this fact in your reporting) there are still good reasons to setup clinical investigations using this approach – such as reduced costs, easier patient enrolment as patients are not asked to do anything extra, a lower regulatory burden and a high real-world generalizability.

However, as SoC data is often less controlled, more heterogeneous, and sometimes indirect compared to device-specific data, this can end up as quite a challenging task.

Additionally, the way SoC is locally implemented can also vary by geographical region, by hospital and even by physician.

Furthermore, there is also the risk of  difference in interpretation of results between the Sponsor / Manufacturer and the physician using the device.  

For example, let’s say that a surgical device is the focal point of the investigation. This device, according to the Instructions for Use, is supposed to be used in a certain way / for certain tasks during surgery (at the surgeon discretion if needed in a particular case). As the Sponsor / Manufacturer, you would most likely set up a study collecting data from the use of the device, and your endpoints would be measuring the success of using the device, or put simply – was the surgeon able to use the device as planned?

If the surgery was not performed as planned, this could mean that as per study endpoints, the case was unsuccessful. However, as per the surgeon, it was successful, since, although steps had to be altered, everything went well in the end.

This example illustrates how the same actual outcomes are interpreted differently depending on the point of view selected.  

To prevent such situations, the Clinical Investigation Plan must provide clear definitions to ensure that all parties have a common understanding of the objectives and endpoints.

If you opt for a more interventional post-market clinical investigation, the normal cause of action is to define the clinical data to be collected to confirm the established clinical performance or effectiveness in a broader population, both when it comes to subjects but also regarding users.

With this setup you can add assessments which are in addition to SoC, whether invasive / burdensome or not (be aware that the regulatory submissions required may differ depending on the assessments you choose to add).

If you decide to go for the observational route, you are dependent on the current clinical practice / SoC, and how it is implemented at each specific study site. As a consequence, you must evaluate SoC per geography/site/physician.

Although there may be official guidelines available, that does not ensure that the SoC process is executed identically at all sites.

There are many potential reasons for such discrepancies, including the way the health insurance system works in the specific region, as well as cultural differences (not to be neglected).

The example illustrates this dilemma: two different sites in two different countries, where at one site the patients in general do not come back for a follow-up visit if not absolutely necessary, and at another site where the patients, in general, like to come back regularly for check-ups.

Thus, it is not safe to assume that the handling, as reported by the lead site, is identically conducted on other sites simply because the guidelines say so, even if all sites are in the same region.

Thus: given that the observational route has been chosen, collecting SoC data, what should you think about?

The strategy for data collection should include the following:

• Minimize the burden: Make use of existing hospital records as much as possible, to avoid "investigator fatigue." Be aware of the fact though, that routine care often has "holes" in documentation compared to clinical trials – this is particularly important if you intend to collect retrospective data.
• Collect quality over quantity: Think carefully through which data is essential to collect in order to reach your objectives and focus on that.
• Timing: Try to align the data collection points with routine clinical follow-up schedules – taking into account that there are sometimes differences in the number of and length of follow-up at different sites. In addition, be aware of the fact that SoC can change during a multi-year PMCF study.

Collecting SoC data can seem like an efficient way, with the least regulatory burden. As seen above the SoC can, however, become a moving target that changes over time.

Strive for avoiding endpoints that cannot realistically be captured in a SoC dataset and do not collect data that you cannot analyze meaningfully.

Although the SoC data may be weaker when it comes to comparisons (and you should be transparent about this fact in your reporting) there are still good reasons to setup clinical investigations using this approach, such as:

• reduced costs
• easier patient enrolment (as patients are not asked to do anything extra)
• a lower regulatory burden and
• a high real-world generalizability.

Carin Nilsson is a senior consultant at confinis. She has supported clinical investigations with medical devices – both as a consultant/CRO, but also on the manufacturer side, since 2006. Her experience includes pre- as well as post-market clinical investigations for class IIa / IIb/ III (including active implantable medical devices) in different geographical regions and including adult as well as a pediatric populations.