EudraVigilance Signal Detection

EudraVigilance is the European Union system for managing and analysing reports of suspected adverse reactions to medicines. For signal management, its value lies not simply in the number of reports it contains but in the ability to examine case-level information and reporting patterns across a very large regulatory dataset.

The role of EudraVigilance in marketing authorisation holder (MAH) signal management changed materially in 2025. Commission Implementing Regulation (EU) 2025/1466 amended Implementing Regulation (EU) No 520/2012 and ended the pilot under which only MAHs responsible for selected active substances were required to perform signal detection in EudraVigilance. The current rule applies to all MAHs with medicinal products authorised in the European Economic Area: they must monitor the data available in EudraVigilance and use those data together with information from other available sources.

At the same time, the former requirement for MAHs to notify EMA and national competent authorities of validated EudraVigilance signals through a standalone signal-notification form was removed. EudraVigilance information now feeds into the MAH's own signal-management process, with resulting actions taken through the applicable regulatory routes such as product-information variations, PSURs and other legally relevant procedures.

This change makes one principle particularly important: EudraVigilance is neither a self-contained signal-management system nor a statistical alarm service. It is a regulatory data source that must be integrated into a multidisciplinary pharmacovigilance process.

Purpose and Regulatory Framework

The purpose of EudraVigilance monitoring is to identify or clarify new risks and changes in known risks that may not be evident from the MAH's internal data alone. The legal basis is principally found in Regulation (EC) No 726/2004, Directive 2001/83/EC and Commission Implementing Regulation (EU) No 520/2012 as amended by Implementing Regulation (EU) 2025/1466.

Article 18(2) of the amended Implementing Regulation now states that MAHs shall monitor data available in EudraVigilance and use those data together with data from other available sources. Article 19 provides that identification of new or changed risks is based on signal detection and analysis and that, for monitoring within EudraVigilance, only signals related to suspected adverse reactions are considered. Article 19 also requires a multidisciplinary approach and statistical analysis where appropriate.

Article 20 describes factors relevant to determining evidentiary value, including clinical relevance, quantitative strength of association, consistency, exposure-response relationship, biological plausibility, experimental findings, analogies and the nature and quality of the data. These requirements make clear that a statistical disproportionality result is only one part of the assessment.

GVP Module IX remains the principal EU guidance on signal management. However, parts of the existing Module IX text pre-date the 2025 legal amendment. EMA has therefore published specific questions and answers on Implementing Regulation (EU) 2025/1466 and has stated that Module IX will be updated to align with the new framework. Where older guidance and the amended legislation differ, the current legislation governs.

What changed after the pilot?

The pilot launched in February 2018 applied the EudraVigilance monitoring obligation to MAHs for a defined list of active substances and combinations. That pilot ended when Implementing Regulation (EU) 2025/1466 entered into force.

The current position can be summarised as follows:

The removal of the standalone notification form does not mean that signals can remain internal indefinitely. The MAH still has the general duty to keep product information current, evaluate safety information, address relevant signals in PSURs where applicable and take other regulatory action when the evidence requires it.

Where EudraVigilance Fits in Signal Management

A useful conceptual sequence is:

all safety sources → detection of a possible issue → EudraVigilance interrogation → validation → prioritisation and assessment → regulatory or risk-management action.

EudraVigilance can enter this sequence at more than one point. An MAH may use it as a primary source during routine screening, but EMA's 2026 Q&A also allows organisations to use EV proportionately at later stages. For example, a signal first identified from internal spontaneous reports or scientific literature may be checked against EudraVigilance during validation to determine whether similar cases exist elsewhere and whether a broader reporting pattern is visible.

During formal assessment, EudraVigilance may provide case forms, line listings and quantitative context that supplement internal cases, clinical-trial data, literature, epidemiology and product-utilisation information. This is especially valuable where the MAH's own case count is small or geographically narrow.

The operating model should therefore answer three questions clearly:

1. Which products and substances are monitored?
2. At what stages and with what frequency is EudraVigilance used?
3. How are EudraVigilance findings integrated with other evidence and documented?

The answers should be defined in the MAH's signal-management procedures rather than inferred retrospectively from individual signal files.

EVDAS as the Analytical Interface

The EudraVigilance Data Analysis System (EVDAS) supports safety monitoring and signal evaluation using EudraVigilance data. EMA describes EVDAS as providing, among other outputs:

• electronic reaction monitoring reports (eRMRs);
• line listings of individual cases; and
• individual case report forms.

EVDAS also includes a disproportionality measure based on the reporting odds ratio (ROR). These outputs help users identify drug-event combinations that deserve medical review and help contextualise signals detected elsewhere.

An eRMR should therefore be understood as an analytical view of the database, not as a list of regulatory signals. A drug-event pair can be statistically disproportionate yet clinically unconvincing, already well characterised, confounded by indication or influenced by stimulated reporting. Conversely, a clinically important signal may exist even when a disproportionality threshold is not crossed, particularly for rare, distinctive or rapidly emerging events.

From Statistical Output to a Safety Question

The first task after an EVDAS finding is not to decide whether the product caused the event. It is to decide whether the observation represents a credible safety question that deserves further work.

A useful review begins with the underlying cases. The reviewer should understand how many unique cases are represented, the seriousness and clinical phenotype, chronology, dose and duration where available, concomitant medicines, competing causes, dechallenge or rechallenge information, relevant investigations and whether the event is already described in the applicable reference information.

The quantitative result should then be interpreted in that clinical context. A reporting odds ratio reflects relative reporting, not disease incidence. It can be affected by differences in reporting behaviour, launch effects, publicity, indication, co-medication, geography, coding practice and the composition of the comparator database.

Signals of disproportionate reporting

A signal of disproportionate reporting (SDR) is a statistical observation generated by a defined signal-detection algorithm. It is not synonymous with a validated safety signal. The distinction matters because the algorithm is deliberately designed to identify observations for review, accepting that many will not survive clinical evaluation.

Thresholds and filters can help manage a large volume of product-event combinations, but they are part of the detection method rather than universal regulatory definitions of a signal. An organisation should therefore document the methodology it uses and understand what the method may miss.

Potential limitations include:

• small-number instability, where a few reports produce a large ratio;
• stimulated reporting, for example after a safety communication;
• confounding by indication, where the underlying disease is associated with the event;
• co-medication confounding, particularly in complex therapeutic settings;
• masking, where heavily reported products or events alter the comparator pattern;
• duplicate or linked reports; and
• coding granularity, where related clinical concepts are split across MedDRA terms.

These limitations do not invalidate disproportionality analysis. They explain why it must be followed by medical review.

A Practical EudraVigilance Review Workflow

The current EU framework does not prescribe one universal screening calendar or a fixed company workflow. EMA's Q&A on Implementing Regulation (EU) 2025/1466 instead expects EudraVigilance use to be proportionate to the product's risk, known safety profile and characteristics and integrated coherently with other safety sources.

A practical workflow can therefore be organised as follows.

1. Define the monitoring strategy

For each product or product group, define how EudraVigilance contributes to signal management. A newly authorised medicine with limited exposure may justify more frequent direct screening than a mature medicine with a stable safety profile, but frequency should arise from documented risk reasoning rather than an invented universal monthly or quarterly rule.

The procedure should also define whether EudraVigilance is used as a primary detection source, during validation of signals found elsewhere, during detailed assessment, or through a combination of these approaches.

2. Retrieve the relevant EVDAS output

The reviewer should retain enough context to reconstruct the analysis. Depending on the question, this may include the eRMR, line listing, individual case forms, date of extraction, product/substance selection, MedDRA terms and relevant filters.

For reproducibility, the organisation should know which dataset and query supported the decision. It is usually unnecessary to preserve every transient screen indefinitely if the controlled process defines what evidence must be retained, but a significant signal decision should not depend on an output that cannot later be reconstructed.

3. Review case-level evidence

Statistical screening prioritises attention; case review determines clinical credibility. The medical reviewer should examine whether the cases describe the same clinical entity, whether temporal relationships are plausible and whether there are recurrent features suggesting a pattern.

A cluster of superficially similar coded terms can prove heterogeneous when narratives are reviewed. Conversely, several differently coded terms can represent one coherent clinical syndrome.

4. Compare with existing knowledge

The finding should be considered against the current product information, RMP where applicable, previous signal assessments, class safety information, clinical-trial experience and known disease background.

A known adverse reaction can still generate an important signal if the new information suggests a change in severity, frequency, outcome, risk factor, mechanism, preventability or affected population.

5. Integrate other sources

EudraVigilance data should be used together with other available sources. Relevant evidence may include:

• the MAH's global safety database;
• scientific and medical literature;
• clinical trials;
• non-interventional studies and registries;
• real-world healthcare data;
• non-clinical or mechanistic findings;
• exposure and utilisation information;
• regulatory actions in other jurisdictions; and
• product-quality information where it has safety implications.

The aim is not to collect every possible source mechanically. The source mix should answer the specific safety question.

6. Validate and prioritise

Signal validation asks whether the available documentation contains sufficient evidence to demonstrate a new potentially causal association or a new aspect of a known association that justifies further analysis.

Prioritisation then considers factors such as seriousness, severity, preventability, public-health impact, vulnerable populations and the potential effect on the benefit-risk balance. The result may be closure with rationale, continued observation, or progression into fuller assessment.

7. Assess and act

A validated signal may require a structured cumulative assessment. Actions depend on the evidence and can include continued monitoring, targeted follow-up, further pharmacoepidemiology, product-information variation, RMP changes, risk minimisation or other regulatory interaction.

Following the 2025 amendment, MAHs no longer submit a standalone validated-signal notification form solely because the signal was detected in EudraVigilance. Instead, the outcome is handled through the appropriate legal and regulatory mechanism.

Relationship With PRAC and Regulatory Signal Management

The MAH's signal-management process and the EU regulatory network are related but not identical. EMA and national competent authorities also monitor EudraVigilance and manage regulatory signals through the processes described in the Implementing Regulation and GVP.

When PRAC or another authority asks an MAH for a cumulative review, analysis or proposed action, EMA expects relevant EudraVigilance data to be considered along with the MAH's other evidence. The company should therefore be able to reproduce the analytical basis of its response and explain differences between company and regulatory case counts or interpretations.

A regulator's signal assessment may result in recommendations for product-information changes, further data generation, risk minimisation or other action. The MAH's responsibility is not merely to await regulatory detection; the company maintains its own obligation to identify, evaluate and act on emerging safety information.

QPPV Oversight

The QPPV does not need to personally review every eRMR or every product-event pair. Effective oversight requires something different: sufficient visibility to determine whether the signal-management system is functioning and whether important issues reach the right decision-makers.

Useful oversight evidence can include trend information on signal workload and timeliness, significant open assessments, overdue activities, major methodological or system changes, serious process deviations and the outcomes of relevant audits or quality reviews. The exact dashboard and escalation thresholds are organisational controls, not prescribed EMA templates.

For a material safety issue, the QPPV should be able to understand how it was detected, what EudraVigilance contributed, what other evidence was considered, what decision was made and whether the resulting regulatory actions were completed.

Special Situations and Interpretation Challenges

Newly authorised or rapidly changing products

Early after launch, case numbers may be small while clinical attention is high. A single serious, distinctive event can be clinically important even without statistical disproportionality. Conversely, publicity and intensive monitoring can amplify reporting. A proportionate strategy therefore combines direct case review with quantitative screening rather than relying on either method alone.

Products with known class effects

An event already associated with the pharmacological class may still require product-specific evaluation. The question is not simply whether the class effect exists, but whether the product shows a different frequency, severity, phenotype, risk factor or mechanism that changes its individual risk profile.

Rare diseases and small populations

For medicines used in small populations, conventional disproportionality may be unstable because both the product and event counts are sparse. Case quality, exposure context, mechanistic plausibility and registry or trial evidence may carry greater weight.

Changes in coding or reporting behaviour

MedDRA version changes, recoding, authority communications, label updates or media attention can alter reporting patterns without an underlying change in biological risk. A sudden quantitative increase should therefore be interpreted against the history of the dataset and the product.

Potential Failure Modes

The following are illustrative failure modes, not cited inspection findings.

The company still operates the old pilot model

A procedure that limits routine EudraVigilance use to the former pilot substance list is obsolete. Since the 2025 amendment, all MAHs with medicines authorised in the EEA fall within Article 18(2).

EVDAS is treated as a separate compliance exercise

Downloading eRMRs on schedule without integrating findings into signal validation and assessment does not demonstrate effective pharmacovigilance. The regulatory requirement concerns monitoring and use of the data, not the production of archived files.

A statistical threshold becomes the definition of a signal

A rigid rule can create both false reassurance and excessive workload. Statistically disproportionate pairs may be non-causal, while clinically important signals may exist below the threshold.

Closure decisions lack a scientific trail

A reviewer may reasonably close an SDR after considering known labelling, duplicate reports, confounding and case quality. The weakness arises when the organisation cannot reconstruct that reasoning.

EudraVigilance and internal data disagree but the difference is ignored

Differences may arise from reporting territories, duplicate management, product coding, ownership of cases, data cut-offs or processing status. Material discrepancies should be understood before conclusions are drawn.

The removal of standalone signal notification is misunderstood

Deletion of Article 21(2) removed one notification mechanism. It did not remove the obligation to assess safety information or to take appropriate regulatory action when the evidence warrants it.

Inspection Considerations

An inspector evaluating EudraVigilance signal detection could reasonably examine whether the MAH has adapted its procedures to the amended legislation and whether the process functions in practice.

Illustrative questions include:

• How did the organisation change its signal procedures after Implementing Regulation (EU) 2025/1466?
• Which medicinal products are covered by EudraVigilance monitoring?
• Where in the signal-management process is EudraVigilance used for each product category?
• How is monitoring frequency justified?
• How are eRMRs, line listings and case forms used and retained?
• How are statistical findings distinguished from validated signals?
• How are signals detected from internal or literature sources checked against EudraVigilance?
• How are significant differences between internal and EudraVigilance data investigated?
• How are signal decisions linked to PSURs, RMPs and product-information changes?
• How does the QPPV obtain visibility of important signals and process failures?

The objective is not to demonstrate that every organisation uses the same schedule or threshold. It is to show that the chosen process is scientifically justified, proportionate, current and traceable.

Practical Review Checklist

A controlled EudraVigilance signal-detection process should allow the organisation to confirm that:

• all EEA-authorised products are considered within the Article 18(2) monitoring framework;
• procedures no longer rely on the terminated pilot list;
• EudraVigilance is explicitly integrated with other safety sources;
• the role and frequency of EV use are proportionate and documented;
• relevant EVDAS outputs can be reproduced or retrieved;
• case-level medical review accompanies quantitative screening where needed;
• ROR and other statistics are treated as screening evidence rather than causal proof;
• signal validation, prioritisation and assessment decisions are documented;
• the removal of the standalone signal-notification form has been reflected correctly in procedures;
• regulatory actions arising from signals use the appropriate legal route;
• important signals can be traced into PSURs, RMPs and product information where applicable;
• methodological, system or access changes are controlled; and
• the QPPV has proportionate oversight of system performance and material safety issues.

Key Takeaways

EudraVigilance is now a mandatory safety-information source for all MAHs with medicinal products authorised in the EEA. The 2025 amendment to Implementing Regulation (EU) No 520/2012 ended the limited MAH pilot and requires EudraVigilance data to be monitored and used together with other available sources.

The same amendment removed the former standalone notification route for MAH-validated EudraVigilance signals. This changes the workflow, not the underlying duty to identify, evaluate and act on safety information.

EVDAS supports that work through eRMRs, line listings, individual case forms and the reporting odds ratio. These are detection and evaluation tools. They do not establish incidence or causality and should not replace multidisciplinary clinical assessment.

A mature process therefore does not ask whether an eRMR was reviewed. It asks whether EudraVigilance information was used at the right point, with the right scientific interpretation, and whether the resulting decision can be traced into the wider pharmacovigilance and regulatory system.

References

1. European Commission. Commission Implementing Regulation (EU) 2025/1466 of 22 July 2025 amending Implementing Regulation (EU) No 520/2012. EUR-Lex. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32025R1466
2. European Commission. Commission Implementing Regulation (EU) No 520/2012, consolidated version current at 12 February 2026. EUR-Lex. https://eur-lex.europa.eu/eli/reg_impl/2012/520/2026-02-12/eng
3. European Medicines Agency. Signal management — monitoring EudraVigilance, legal basis and termination of the MAH signal-detection pilot. https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/signal-management
4. European Medicines Agency. Questions and answers on Implementing Regulation (EU) 2025/1466: Amendment of Regulation (EU) No 520/2012 and Conclusion of the Signal Detection in EudraVigilance Pilot by MAHs, EMA/243145/2025 Rev. 2, updated 28 January 2026. https://www.ema.europa.eu/system/files/documents/other/questions-answers-implementing-regulation-eu-2025-1466-en.pdf
5. European Medicines Agency. EudraVigilance system overview — EVDAS, eRMR, line listings, individual case forms and reporting odds ratio. https://www.ema.europa.eu/en/human-regulatory-overview/research-development/pharmacovigilance-research-development/eudravigilance/eudravigilance-system-overview
6. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP) Module IX – Signal management (Rev. 1) and Addendum I. Available from the EMA GVP collection. https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/good-pharmacovigilance-practices-gvp
7. European Medicines Agency. 2025 Annual Report on EudraVigilance for the European Parliament, the Council and the Commission. https://www.ema.europa.eu/en/documents/report/2025-annual-report-eudravigilance-european-parliament-council-commission_en.pdf

Regulatory Note

Article 18(2) of Implementing Regulation (EU) No 520/2012, as amended by Implementing Regulation (EU) 2025/1466, is legally binding within its scope. EMA's January 2026 Q&A provides current practical interpretation of the amended requirements while GVP Module IX is being aligned with the new legal framework. EVDAS review schedules, internal thresholds, dashboards, retention models and inspection questions described in this article are recommended or illustrative practices unless an authoritative requirement is specifically identified. Statistical disproportionality is a signal-detection aid and should not be presented as proof of causality or incidence.