Post-market clinical follow-up (PMCF) surveys are widely used. They are not widely understood. Manufacturers reach for them because they look manageable, they can be run without ethics committee involvement in many cases, and they produce data fast. What they often fail to account for is that the EU Medical Device Regulation (EU MDR) has a very specific view of what counts as credible clinical evidence, and a badly designed survey does not get manufacturers any closer to it.
What follows is a practical breakdown of where surveys fit in the PMCF evidence hierarchy, where they fall short, what regulators and notified bodies actually expect, and how to design one that holds up.
Under the EU MDR, manufacturers must implement a proactive system for collecting and evaluating post-market data throughout the entire lifecycle of a medical device. That data feeds into clinical evaluation updates, risk management, post-market surveillance documentation, and the summary of safety and clinical performance (SSCP). The goal is continuous confirmation that the device's benefit-risk profile remains acceptable.
MDCG 2020-6, Appendix III provides a hierarchy of clinical evidence sources ranked from strongest to weakest across 12 levels. Where a PMCF survey lands on that scale depends entirely on how it is designed.
A general survey with a retrospective design, based on recalled experiences and unverified data, typically provides Level 8 evidence out of 12. Level 8 is adequate for general post-market surveillance activities. For higher-risk devices, it falls well short of what notified bodies expect as high-quality clinical data.
A Level 4 survey is a different category entirely. Level 4 surveys are case-based, structured under clinical and regulatory supervision, include a statistical sample size justification, and use validated tools that ensure data traceability and integrity. Level 4 is the minimum evidence level that notified bodies typically expect for PMCF evidence on Class III, implantable, and non-well-established technology legacy devices.
The distinction matters because the same word, "survey," can refer to activities at very different levels of rigor and regulatory value.
PMCF surveys work well for many post-market clinical follow-up objectives. They are particularly well-suited for capturing real-world use information, clinician feedback, usability insights, and patient-reported outcomes from the people who actually use the device in clinical settings. When a device moves from a controlled clinical investigation environment into routine clinical practice, new patterns emerge. Conditions are less controlled, patient populations are broader, and less experienced clinicians may be using the device. Surveys can capture what happens in that real-world context at scale.
Long-term observational follow-up is another strong use case. Manufacturers can run open survey tools that collect data continuously and evaluate results at defined intervals to identify safety and performance trends. This is exactly the type of ongoing monitoring that the MDR envisions under Annex XIV Part B.
Patient and caregiver experience surveys provide increasingly relevant data. Capturing patient-reported outcomes or treatment experience generates evidence relevant to both regulatory requirements and value-based healthcare discussions.
When designed correctly, PMCF survey data can directly support clinical evaluation updates, risk management, and the SSCP. For more on selecting the right PMCF method for your device, the guide to selecting the ideal PMCF activity is worth reading alongside this article.
FREE DOWNLOAD: Click here to download your free 15-in-1 Clinical Investigations Content Bundle.
The limitations are real and worth understanding clearly before committing to a survey-based approach.
PMCF surveys become problematic when the questions move beyond the approved intended purpose of the device. If a survey is designed to explore potential new indications, validate off-label uses, or support new clinical claims, the activity may cross into clinical investigation territory under MDR Articles 62 through 81. MDCG 2020-7 and MDR Annex XIV Part B both emphasize that PMCF activities should focus on confirming safety and performance within the approved intended purpose. Running a survey to identify and validate off-label use is particularly risky. It can be interpreted as the manufacturer allowing or promoting off-label use, and if off-label use data surfaces during the survey, the manufacturer has obligations under MDR Annex XIV Part B, 6.1(e) to assess risk implications and potentially update labeling.
Surveys also have limited value when the goal is to collect objective clinical performance data or safety endpoints that require clinical examination beyond normal clinical practice. If the investigation requires burdensome procedures for the patient, a formal clinical study with ethics oversight and regulatory notification is the right path.
The question comes up often in practice. The boundary is worth understanding clearly.
A clinical investigation under MDR Articles 62 through 81 and Annex XV involves a systematic prospective study with a predefined protocol, typically requires competent authority notification and ethics committee approval, may involve additional procedures beyond normal clinical use, and follows the full ISO 14155 framework.
A Level 4 PMCF survey is also case-based and protocol-driven. The key distinction under MDR Article 74 is whether subjects are submitted to additional invasive or burdensome procedures, and more generally, the type of data being collected. A Level 4 survey collects data within the approved intended purpose under normal conditions of use, without imposing additional procedures. It uses structured questionnaires completed by clinicians using patient records.
A Level 4 PMCF survey should still follow applicable good clinical practice per ISO 14155 for data collection methodology to be credible to notified bodies. If a manufacturer's data collection request is structured as a systematic prospective study with predefined endpoints and burdensome procedures, the honest answer is that it is a clinical investigation, not a PMCF survey, and it should proceed accordingly.
For a deeper comparison, the article on clinical evaluation versus clinical investigation covers the frameworks in detail.
Under the EU MDR, there is no predefined sample size for PMCF activities. Manufacturers must provide a scientific justification explaining why the chosen sample size is appropriate given the device type, the associated risk, and the objectives of the data collection.
MDCG 2020-7, Section C explicitly warns against PMCF surveys with no statistical justification. Several factors determine what an appropriate sample size looks like in practice.
Device risk class plays a role. Higher-risk devices generally require broader and more strong data sets. Lower-risk devices may rely on smaller but still scientifically defensible samples.
The type of outcome being evaluated matters. Objective clinical performance endpoints usually require larger data sets, whereas usability feedback, user experience, or patient-reported outcomes may require fewer participants, provided the methodology minimizes bias.
Event rate is critical for safety monitoring. If the objective is to detect rare adverse events, large populations or longer follow-up periods may be necessary. The Rule of Three is a useful reference point: with zero events observed in n subjects, the upper 95% confidence limit for the true event rate is approximately 3/n. For example, 385 participants with zero adverse events gives an upper bound of roughly 0.78%. Whether that is adequate depends on the expected frequency and clinical significance of the adverse event.
The focus under the EU MDR is on scientific validity and the ability to confirm safety and performance under real-world conditions. Regulators typically expect a clear methodological rationale, appropriate statistical consideration, sufficient data to detect meaningful trends, and transparency about the limitations of the data set. For guidance on structuring the broader PMCF plan, the PMCF plan development guide covers this in detail.
PMCF surveys can be fully anonymized, and in many situations they should be designed that way from the outset. The appropriate approach depends on who the participants are and whether longitudinal follow-up is required.
For patients and lay users, full anonymization is typically the safest approach. Questionnaires should focus on device experience, usability, safety signals, and patient-reported outcomes while avoiding unnecessary personal health information. This aligns with the GDPR principle of data minimization.
When longitudinal follow-up is required, full anonymization becomes more difficult because responses must be linked across multiple time points. Pseudonymization is the standard approach in those situations. The identifying information is managed locally by the investigator or healthcare provider, while the manufacturer receives only de-identified data. This allows longitudinal analysis while maintaining strong privacy protection.
For healthcare professionals responding to surveys in their professional capacity, the privacy risk is generally lower. Their feedback can often be collected without sensitive personal data, though GDPR principles and data minimization requirements still apply.
Survey fatigue is a real and growing problem. Users are overwhelmed with survey requests, and even well-designed surveys get ignored when response rates drop. Strategies that help include keeping surveys as short as possible, using branch logic to avoid redundant questions, communicating the regulatory purpose clearly, timing surveys around clinical encounters, and engaging key opinion leaders at clinical sites.
More broadly, manufacturers should not rely solely on PMCF surveys. MDCG 2020-6 Appendix III offers 12 evidence levels, and MDCG 2020-7 lists multiple PMCF activity types. A diversified approach is more sustainable and more credible than a single-method strategy.
PMCF is a continuous process under MDR Annex XIV Part B, but continuous does not mean the same survey runs indefinitely. PMCF surveys can be conducted in defined cycles aligned with the PMCF plan, and the type and intensity can evolve over time. A Level 4 survey might run in years one and two, then transition to Level 8 surveys plus literature monitoring if results are satisfactory. For Class III and implantable devices, PMCF reports must be updated at least annually per MDR Article 61(11).
Manufacturers should also consider engaging in structured dialogue with their notified body about PMCF plans. These conversations can align expectations early and prevent costly misalignment later in the certification process. There is sometimes a tendency to avoid them, but it is entirely feasible to present a clinical development plan and discuss the expected approach before committing to a methodology.
The structure is ultimately what distinguishes a PMCF survey that regulators will accept from one they will not. A survey protocol, defined endpoints, a justified sample size, validated collection tools, and a clear analysis plan are the foundation. Plan carefully, collect rigorously, and report transparently.
If you are building out your PMCF data collection process, these related guides go deeper on the specific components:
Greenlight Guru Clinical is an electronic data capture (EDC) platform purpose-built for medical device clinical data collection, including PMCF surveys. See how it supports PMCF data collection and EU MDR compliance.