Why cardiovascular device CERs fail at the safety endpoint

Written by HATEM RABEH, MD, MSc Ing

Your Clinical Evaluation Expert And Partner

in
S

This is not an isolated case. Cardiovascular devices live in a regulatory space where the consequences of failure are immediate and life-threatening. Yet many clinical evaluation reports approach safety assessment with the same methods used for lower-risk devices. The gap between what regulators expect and what manufacturers deliver is not about missing data. It is about misunderstanding what safety means in this context.

The temporal dimension of cardiovascular risk

Cardiovascular devices interact with dynamic biological systems. A vascular stent does not simply succeed or fail at implantation. It creates a cascade of risks that evolve: acute thrombosis in the first 24 hours, subacute thrombosis in the first month, late stent thrombosis beyond a year, neoatherosclerosis over multiple years.

Most CERs I review present adverse event tables that aggregate events across the entire follow-up period. They report percentages: 2.3% device-related serious adverse events. This tells you almost nothing about risk.

The critical question is not how many events occurred. It is when they occurred, under what conditions, and whether the risk profile changes over time. A 2% thrombosis rate concentrated in the first 30 days has completely different clinical implications than the same rate distributed evenly over five years.

The equivalence trap in cardiovascular claims

Equivalence claims for cardiovascular devices routinely fail because manufacturers mistake technical similarity for clinical equivalence. I see this pattern repeatedly: same material platform, same delivery system, similar radial force characteristics. The manufacturer concludes equivalence and relies heavily on literature data from the comparator device.

Then the questions arrive from the Notified Body. Why does your device have a different strut thickness? How does this affect endothelialization kinetics? What is the impact on flow dynamics at bifurcation points? Can you demonstrate that inflammatory response profiles are comparable during the critical first 90 days?

These are not theoretical concerns. Strut thickness differences of 30 micrometers have shown measurable impacts on neointimal hyperplasia. Drug release kinetics that differ by days can alter the local inflammatory environment enough to change thrombosis risk.

The technical and biological characteristics defined in Annex XIV of the MDR are not a checklist. They are a framework for understanding what actually drives cardiovascular device risk. For implantable cardiovascular devices, you cannot simply claim equivalence and lean on literature. You must demonstrate that the specific characteristics affecting biocompatibility, thrombogenicity, mechanical interaction with tissue, and temporal risk profile are sufficiently similar.

What safety endpoints actually mean in this space

Cardiovascular clinical evaluation demands clarity about what constitutes a safety endpoint and why it matters. I have reviewed CERs that list “no device-related deaths” as a primary safety conclusion. This is insufficient.

Death is an endpoint, but it is often too late and too insensitive for cardiovascular devices. The meaningful safety endpoints are the precursors: target lesion revascularization, stent thrombosis, myocardial infarction attributable to device failure, major bleeding requiring intervention, vascular complications at access sites.

Each of these endpoints has diagnostic criteria. Stent thrombosis, for instance, follows Academic Research Consortium definitions: definite, probable, possible. Your clinical evaluation must specify which definition you use, why, and how your data collection ensures reliable classification. If your PMCF plan collects “thrombosis events” without specifying ARC criteria and timing windows, you have not defined your endpoint.

This matters during review because regulators and Notified Bodies assess whether your safety conclusions are based on endpoints that can actually detect the risks your device creates. Vague endpoints produce vague conclusions. Vague conclusions fail review.

The state of the art for cardiovascular devices

SOTA analysis for cardiovascular devices is where many CERs show their weaknesses most clearly. The manufacturer reviews current guidelines, summarizes published meta-analyses, and concludes that their device meets or exceeds current performance standards.

But SOTA is not about showing your device is acceptable. It is about demonstrating you understand what the current generation of devices achieves and where the field is moving. For drug-eluting stents, SOTA includes understanding that newer-generation devices have reduced very late stent thrombosis rates compared to earlier generations. It includes knowing that biodegradable polymer platforms are changing the risk-benefit calculation. It includes recognizing that dual antiplatelet therapy duration recommendations have evolved based on device-specific thrombosis data.

Your CER must position your device within this evolving landscape. If your device uses a durable polymer when the field is moving toward biodegradable options, you must address why this choice is justified. If your recommended DAPT duration differs from current guidelines for similar devices, you must provide evidence supporting this recommendation.

Reviewers read SOTA sections to see whether you know your field. They can identify immediately when a SOTA section is assembled from abstracts rather than constructed from deep understanding of how cardiovascular device safety has evolved over the past decade.

Building a safety-first clinical evaluation structure

A safety-focused CER for a cardiovascular device starts with hazard identification that is genuinely specific. Not generic lists from ISO 14971 templates. Specific failure modes for your device: what happens if your polymer coating degrades faster than expected? What are the consequences if your deployment mechanism creates edge dissection at a bifurcation? What occurs if your radiopaque markers migrate?

Each identified hazard must connect to a safety endpoint in your clinical data. If you identify “incomplete apposition” as a hazard but your clinical studies do not assess apposition through imaging follow-up, you have not evaluated the safety risk. The gap will be identified during review.

The clinical data section must present safety results with appropriate time stratification. Early safety (0-30 days), mid-term safety (1-12 months), and long-term safety (beyond 1 year) are not arbitrary divisions. They correspond to different biological processes and different risk mechanisms.

For each time period, your analysis must address whether observed event rates align with predicted hazard frequency. If your risk management file estimates 1.5% acute thrombosis risk but your clinical data shows 0.2%, you must address this discrepancy. It might mean your risk estimation was conservative. It might mean your follow-up was incomplete. It might mean your event detection was inadequate. Unexplained discrepancies between predicted and observed risk undermine reviewer confidence.

PMCF design for cardiovascular devices

PMCF plans for cardiovascular devices cannot be generic registry protocols. They must be designed to answer specific safety questions that remain after pre-market clinical investigation.

I frequently see PMCF plans that propose to collect “real-world safety data” without specifying what safety questions remain unanswered. This is insufficient. Your PMCF plan must identify knowledge gaps: Are very late thrombosis rates in your device comparable to equivalent devices beyond three years? Does your device perform safely in heavily calcified lesions when used off-label? What is the reintervention rate in small vessel diameters?

Each knowledge gap should connect to a PMCF objective, which connects to specific endpoints, which connects to sample size justification and follow-up duration. If you cannot draw this chain of logic clearly in your CER, your PMCF plan is not adequately justified.

Reviewers will also assess whether your PMCF follow-up duration matches the temporal risk profile of your device. A permanent implant with known late complications requires long-term follow-up. Proposing 12-month PMCF follow-up for a device with known risks of very late stent thrombosis signals that you have not understood your own safety profile.

How reviewers actually assess cardiovascular device CERs

When a Notified Body reviews a cardiovascular device CER, they are not checking boxes. They are assessing whether you demonstrate understanding of cardiovascular risk mechanisms and whether your evidence addresses those mechanisms.

They look at your patient population. If your clinical study enrolled only straightforward lesions but your IFU allows use in complex anatomy, they will identify this gap. If your study excluded patients on novel anticoagulants but these patients will use your device post-market, the risk evaluation is incomplete.

They examine your follow-up completeness. Lost to follow-up rates above 10% at one year raise concerns. If patients with adverse events are more likely to be lost to follow-up, your safety data is biased. If you cannot demonstrate that lost patients have similar baseline characteristics to completed patients, your conclusions are questionable.

They assess your event adjudication process. Were events classified by independent reviewers? Were imaging studies evaluated by core labs? Were clinical events adjudicated by cardiologists not involved in the procedures? Without independent adjudication, cardiovascular event data lacks credibility.

These are not formalities. They reflect whether your evidence can support the safety claims you make in the CER. Weak evidence produces weak conclusions, which produce additional questions, which delay certification.

Connecting safety evidence to benefit-risk conclusions

The benefit-risk analysis in cardiovascular device CERs often feels like an afterthought. It appears as the final section, briefly states that benefits outweigh risks, and moves to conclusions.

This approach misses the point. Benefit-risk analysis for cardiovascular devices is a structured evaluation of trade-offs. Your device might reduce restenosis but increase early thrombosis. It might improve deliverability but increase vascular access complications. It might extend durability but require longer antiplatelet therapy.

The analysis must acknowledge these trade-offs explicitly and demonstrate why the balance favors your device for the intended patient population. If your device is indicated for high-risk patients with few alternatives, the acceptable risk threshold differs from devices intended for stable patients with multiple treatment options.

Reviewers evaluate whether your benefit-risk reasoning matches clinical reality. If you claim benefits based on surrogate endpoints but risks based on clinical events, the comparison is invalid. If you compare your device’s benefits to no treatment rather than to alternative treatments, you have not demonstrated clinical value.

The benefit-risk section should be the core of your CER, not an appendix. It is where you demonstrate that you understand what your device achieves, what it costs in terms of risk, and why that trade-off makes clinical sense.

What happens when safety evaluation is insufficient

The consequence of inadequate safety evaluation is not simply a request for additional information. It is a fundamental question about whether sufficient evidence exists to support certification.

I have seen certification processes stall for 18 months because the safety evidence could not support the intended use claims. The manufacturer had data, but the data did not address the right questions. Additional studies were required. The cost in time and resources was substantial.

More concerning is the post-market consequence. A device certified with inadequate safety evaluation enters the market without proper risk characterization. When adverse events occur, the manufacturer lacks baseline data to assess whether rates are acceptable. PMCF becomes reactive rather than systematic. Regulatory authorities may impose additional restrictions or suspend certification.

The path forward is not more data for the sake of data. It is structured thinking about what cardiovascular safety means for your specific device, which risks matter most, and how your evidence addresses those risks with appropriate temporal resolution and clinical relevance.

Safety-focused clinical evaluation is not a regulatory burden. It is the foundation for understanding whether your cardiovascular device works as intended when it matters most—when patient lives depend on it.

Peace,
Hatem
Clinical Evaluation Expert for Medical Devices
Follow me for more insights and practical advice.