When Three Complaints Become a Regulatory Emergency

Written by HATEM RABEH, MD, MSc Ing

Your Clinical Evaluation Expert And Partner

in
S

The MDR requires manufacturers to report serious incidents and field safety corrective actions. That part is understood. What is not understood is the obligation to report trends even when no single incident meets the threshold of a serious incident.

I have reviewed PMS systems that capture complaints, analyze them, and close them individually. But trends are not addressed as events. They are treated as risk signals for internal monitoring. That disconnect causes silent non-compliance.

The consequence is simple. A Notified Body reviews your complaints database during surveillance. They identify three similar device failures over six months. None were reported. You explain that none qualified as serious incidents. They respond that the trend itself was reportable under MDR Article 88. The audit finding follows.

The MDR Obligation on Trend Reporting

MDR Article 88 establishes the obligation to report serious incidents and field safety corrective actions. But it also includes an often-overlooked requirement. Manufacturers must report any significant increase in the frequency or severity of incidents that are not serious but that might become so.

This means that even if individual complaints do not meet the criteria for a serious incident, a pattern of similar complaints can trigger a reporting obligation.

The language is deliberate. It shifts responsibility from reacting to serious harm to anticipating it. The regulation assumes that trends precede failures. The expectation is that manufacturers detect and act before harm escalates.

But here is the challenge. The regulation does not define what constitutes a reportable trend. It does not specify thresholds. It does not provide formulas. The manufacturer is expected to assess whether a pattern is significant.

That ambiguity is intentional. It places judgment in the hands of those who know the device best. But it also creates confusion and inconsistency.

Why Trends Are Missed

Most complaints management systems are designed to process incidents individually. A complaint comes in. It is categorized. It is investigated. It is closed. If it does not meet serious incident criteria, it remains in the database as a closed record.

That structure works for single-event analysis. But it breaks down when trends need to be identified across time and product batches.

I see three common gaps:

First, no formal trend analysis intervals. The system captures data but does not systematically review patterns weekly, monthly, or quarterly. Trends emerge only when someone manually notices them or when a Notified Body asks the question.

Second, no definition of what a trend is. Some manufacturers say three similar complaints. Others say five. Others use statistical thresholds. Without a documented rule, trend identification becomes subjective and inconsistent.

Third, no link between trend detection and vigilance reporting. Even when a trend is identified, it is handled as a risk management action, not as a potential reportable event. The PMS report notes the trend. The risk file is updated. But no one evaluates whether it meets MDR Article 88 obligations.

What Makes a Trend Reportable

The question every manufacturer must answer is this: when does a pattern of non-serious incidents become a reportable trend?

There is no universal answer. But the decision process should be documented and consistently applied.

Start with the nature of the complaint. Are the incidents related to the same failure mode? Are they linked to the same lot or production period? Are they occurring in a specific user group or clinical setting?

Then consider the frequency. Has the rate of similar complaints increased compared to historical data? Is the increase statistically significant or just expected variation?

Next, evaluate the potential severity. Even if none of the incidents caused harm, could the failure mode escalate under different conditions? Could it affect a vulnerable population or a critical clinical use?

Finally, assess the adequacy of current risk controls. If the trend suggests that existing mitigations are not effective, that is a signal that the risk is not under control.

This is not a checklist. It is a reasoning process. And it must be documented.

The Role of Statistical Thresholds

Some manufacturers implement statistical thresholds. For example, a complaint rate that exceeds two standard deviations above the historical mean triggers trend analysis.

That approach provides objectivity. But it must be calibrated to the device and the complaint type. A high-volume, low-risk device may have different thresholds than a low-volume, high-risk implant.

The threshold should also account for expected variation. Seasonal factors, changes in user base, and product modifications can all affect complaint rates. A spike may reflect normal variation rather than a systemic issue.

But here is the critical point. A statistical threshold can support decision-making, but it cannot replace clinical judgment. The manufacturer must still evaluate whether the trend represents a safety concern.

How Reviewers Assess Trend Reporting

During surveillance audits, Notified Bodies and competent authorities review complaints databases. They do not just check whether serious incidents were reported. They look for patterns that should have triggered trend reporting.

They typically start by filtering complaints by type, failure mode, and time period. If they identify a cluster of similar complaints, they ask whether the manufacturer detected it, analyzed it, and considered it for reporting.

If the answer is no, the finding is immediate. If the answer is yes but the manufacturer decided not to report, the reviewer will ask for the justification. That justification must be documented and reasoned.

I have seen audits where the manufacturer had solid trend detection but weak documentation of the decision process. The trend was noted in a PMS meeting. The team discussed it. They concluded it was not reportable. But there was no written rationale. The Notified Body flagged it as insufficient evidence.

Building a Trend Reporting System

A functional trend reporting system requires four elements.

First, defined review intervals. Trend analysis should happen at regular intervals. Monthly is common for high-volume devices. Quarterly may be sufficient for low-volume implants. The interval should be documented in the PMS plan.

Second, clear trend criteria. Document what constitutes a trend. Is it three similar complaints in a month? Is it a statistically significant increase? Is it any failure mode linked to a specific lot? The criteria should be tailored to the device but must be consistent.

Third, a documented evaluation process. When a trend is detected, document the evaluation. What is the failure mode? What is the frequency? What is the potential severity? What are the current risk controls? Is the trend reportable under MDR Article 88? The answers must be written and signed.

Fourth, integration with vigilance procedures. Trend evaluation should not sit separately from incident reporting. It should be part of the same workflow. When a trend is flagged, it should automatically trigger a vigilance assessment.

Documenting the Decision

The documentation does not need to be complex. A simple trend analysis form works. It should capture the detected pattern, the evaluation criteria, the risk assessment, and the conclusion.

If the trend is not reportable, explain why. If risk controls are adequate and the failure mode is well understood, say so. If the frequency is within expected variation, document the comparison to historical data.

If the trend is reportable, document the notification timeline and the rationale for the timing.

This record is what Notified Bodies and competent authorities will review. It demonstrates that the manufacturer is actively monitoring and reasoning, not just reacting.

The Intersection with PMCF and Risk Management

Trend analysis is not isolated. It connects to PMCF and risk management.

If a trend is detected, it should feed into the clinical evaluation. Is the failure mode consistent with the risk analysis? Does it suggest that a residual risk is higher than predicted? Does it require updating the clinical evaluation report?

It should also inform PMCF activities. If a trend suggests an emerging safety concern, the PMCF plan may need to include targeted data collection or literature monitoring on that specific issue.

And it must update the risk management file. A trend that does not meet reporting thresholds today may still require revised risk controls or updated warnings.

The point is that trend detection is not just a vigilance obligation. It is a feedback loop that keeps the entire post-market system responsive.

Final Thought

Trend reporting is not about reacting to harm. It is about anticipating it. The challenge is that trends are subtle. They do not announce themselves. They require active detection, consistent criteria, and reasoned judgment.

Most manufacturers have the data. They capture complaints. They analyze incidents. But the step from pattern to reportable event is often missing. That gap is where non-compliance hides.

The solution is not a new software system. It is a documented process that connects detection, evaluation, and decision-making. And it requires that the team understands that some patterns are not just signals. They are events.

In the next post, we will look at how to design PMCF activities that actually capture the data you need for trend analysis and clinical evaluation updates. Because trend reporting relies on the quality of the data you collect.

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