QPPV.COM

How to Read an EMA Referral From a Pharmacovigilance Perspective

How to reconstruct the pharmacovigilance reasoning behind an EMA referral, from the initial safety concern and evidence assessment through PRAC recommendations, regulatory decisions and implementation in product information.

Audio Lesson 14 min
Knowledge Assessment Test your understanding of this article. Take the assessment →

How to Read an EMA Referral From a Pharmacovigilance Perspective

Introduction

An EMA referral is one of the most useful publicly available sources for understanding how pharmacovigilance evidence is translated into regulatory action.

It is also easy to read incorrectly.

A referral page may contain a notification, scientific background, a list of questions, procedural documents, a PRAC assessment report, a PRAC recommendation, a CHMP opinion or CMDh position, annexes containing revised product information and, where applicable, a European Commission decision.

These documents do not have the same purpose.

The notification explains why the procedure was initiated.

The list of questions identifies the scientific and regulatory issues that regulators wanted addressed.

The assessment report explains how the available evidence was considered.

The PRAC recommendation records the pharmacovigilance committee's conclusion.

The subsequent committee opinion or position forms part of the applicable regulatory pathway.

The final legal decision, where applicable, establishes the regulatory outcome.

For a pharmacovigilance professional, the useful question is therefore not simply:

What did EMA decide?

It is:

How did the available evidence lead from the original concern to the regulatory conclusion?

That requires the referral to be read as a chain of scientific and regulatory reasoning.

This article provides a structured method for doing so.

It focuses particularly on pharmacovigilance referrals involving safety concerns, including Article 31 referrals arising from pharmacovigilance activities. The exact procedural route depends on the legal basis, authorisation status and products involved, and those elements should always be verified for the individual procedure.

Learning Objectives

After reading this article, the reader should be able to:

1. What Is an EMA Referral?

An EMA referral is a formal regulatory procedure used to obtain an EU-level assessment of a medicinal product or group of medicinal products and, where applicable, a harmonised regulatory outcome.

There are several legal bases for referral procedures.

For pharmacovigilance professionals, Article 31 of Directive 2001/83/EC is particularly important because it provides a mechanism for EU-level consideration of concerns relating to authorised medicinal products where the interests of the Union are involved.

An Article 31 pharmacovigilance referral applies when the procedure is initiated as a result of the evaluation of pharmacovigilance data concerning authorised medicinal products.

Where an Article 31 procedure concerns pharmacovigilance, the scientific assessment is led by the Pharmacovigilance Risk Assessment Committee (PRAC).

The resulting recommendation then enters the applicable regulatory pathway. Depending on the products and authorisation route, this may involve:

The exact route must be established from the individual referral.

The first rule when reading an EMA referral is therefore:

Identify the procedure before interpreting the safety conclusion.

2. A Referral Is a Process, Not a Safety Conclusion

The initiation of a referral does not itself establish causality.

A referral may begin because there is:

The referral creates a formal process through which those questions can be assessed.

This distinction matters because the language used at the beginning of a procedure may be substantially more uncertain than the language used in the final regulatory decision.

A referral can therefore be understood as:

a structured investigation and regulatory decision-making process arising from a defined concern.

3. Start With the Procedure

Before reading the scientific evidence, identify:

  1. the legal basis;
  2. the initiating authority or Member State;
  3. the products involved;
  4. the active substance or substances;
  5. the authorisation status;
  6. the date the procedure started;
  7. the committee responsible for the scientific assessment;
  8. the subsequent regulatory pathway.

This establishes the context in which every later document must be interpreted.

For example, a statement made by PRAC should not automatically be described as a final legal decision.

Similarly, an EMA public summary should not be treated as though it were the complete scientific assessment.

4. The Basic Regulatory Architecture

A simplified model for a pharmacovigilance referral is:

Safety concern
      |
      v
Initiating information / referral
      |
      v
Formal referral procedure
      |
      v
Scientific questions
      |
      v
Evidence submitted and assessed
      |
      v
PRAC scientific assessment
      |
      v
PRAC recommendation
      |
      +----------------------+
      |                      |
      v                      v
     CHMP                   CMDh
      |                      |
      +----------+-----------+
                 |
                 v
       European Commission
          where applicable
                 |
                 v
        Regulatory outcome
                 |
                 v
    Product information / risk minimisation

This is a conceptual model rather than a representation of every possible referral pathway.

The key distinction is between:

scientific assessment

and

regulatory decision-making.

5. The Referral Page Is the Map

The EMA referral page should normally be the first document reviewed.

It establishes:

The page may contain documents from several stages of the procedure.

Do not assume that the most recently published document is necessarily the most useful one for understanding the original scientific question.

A historical reading requires the documents to be placed in chronological order.

6. Which Documents Matter Most?

A useful reading order is:

  1. referral overview;
  2. notification or initiating document;
  3. scientific background, where available;
  4. list of questions;
  5. timetable or procedural history;
  6. PRAC assessment report;
  7. PRAC recommendation;
  8. CHMP opinion or CMDh position, where applicable;
  9. European Commission decision, where applicable;
  10. revised product information and other annexes.

Not every procedure will contain exactly this set of documents.

The principle is to distinguish:

why the procedure started

from

how the evidence was assessed

from

what regulatory action resulted.

7. The Notification: What Triggered the Procedure?

The notification is often the most useful document for understanding the initial state of knowledge.

Ask:

The notification should be treated as the starting point of the regulatory question, not as the answer.

This distinction is particularly important in historical analyses.

Later evidence should not be projected backwards into the initial stage of the procedure.

8. The Scientific Background

Where available, the scientific background may provide the evidence that prompted the referral.

It can help identify:

The important question is:

What was actually known before the formal EU-level assessment began?

This provides the baseline against which the later assessment can be understood.

9. Find the Regulatory Question

Before reading the assessment report in detail, identify the question that the regulators were trying to answer.

It may concern:

These are different questions.

For example:

Does the medicine cause the event?

is different from:

Are the existing risk-minimisation measures sufficient to prevent the event?

The first is principally a causal-assessment question.

The second is principally a risk-management question.

A referral may address both.

10. The List of Questions

The list of questions is one of the most informative documents on a referral page.

It shows what the regulators considered important enough to require formal assessment.

Questions may request analysis of:

The questions can therefore be converted into an analytical framework.

Regulatory question Evidence that may address it
Is there an association? Cases, clinical studies, epidemiology
How large is the risk? Exposure and comparative data
Who is most vulnerable? Subgroup and risk-factor analyses
Does dose matter? Dose-response and exposure-response analyses
Can the risk be prevented? Risk-management and effectiveness evidence
Are existing measures sufficient? Implementation and effectiveness data
Does benefit-risk remain favourable? Integrated benefit-risk assessment

This is often a much better starting point than reading the assessment report linearly.

11. PRAC and the Scientific Assessment

PRAC is the committee responsible for the pharmacovigilance assessment in the relevant procedures.

The scientific assessment is broader than determining whether a statistical association exists.

PRAC may consider:

The final assessment therefore represents an integration of evidence rather than a single statistical test.

12. Read the Assessment Report as an Argument

A long assessment report becomes easier to understand if it is treated as an argument with five components.

1. Question

What exactly is being investigated?

2. Evidence

What data are available?

3. Interpretation

What do those data indicate?

4. Uncertainty

What remains unresolved?

5. Consequence

What regulatory action follows?

When reading, continually ask:

What is the evidence for this statement?

and:

What evidence argues against it?

That discipline prevents the reader from simply accepting conclusions without understanding their basis.

13. Evaluate Each Evidence Stream Separately

Evidence from different sources should not be treated as interchangeable.

13.1 Individual Case Reports

For individual cases, examine:

The number of cases matters.

The quality and consistency of the cases matter more than a simple count.

13.2 Spontaneous Reports

Consider:

A spontaneous reporting system generally does not provide a complete denominator.

Therefore:

A report count should not automatically be interpreted as incidence.

13.3 Clinical Trials

Assess:

Clinical trials can provide strong comparative evidence.

However, they may be poorly powered to detect very rare events.

A negative trial therefore does not necessarily exclude a rare serious risk.

13.4 Epidemiological Studies

Examine:

A statistically significant association is not automatically causal.

A non-significant association does not necessarily demonstrate absence of risk.

13.5 Literature

Assess:

The number of publications is not itself a measure of evidence strength.

13.6 Pharmacological and Non-Clinical Evidence

Mechanistic evidence can contribute to biological plausibility.

Examples include:

Mechanistic evidence should be distinguished from clinical evidence.

A plausible mechanism can strengthen a causal hypothesis.

It does not by itself establish the frequency or clinical significance of the adverse event.

14. Look for Convergence

Strong regulatory assessments often depend on convergence across evidence sources.

For example:

Consistent clinical phenotype
          +
Compatible temporal relationship
          +
Epidemiological association
          +
Dose-response
          +
Biological plausibility
          |
          v
Greater confidence in the association

The converse is equally important.

If:

the totality of evidence may be substantially weaker than an isolated positive finding suggests.

15. Look for Evidence Against the Safety Hypothesis

A strong assessment should consider evidence both for and against the concern.

Look specifically for:

The purpose is not to make the evidence artificially symmetrical.

It is to understand how the regulator weighed conflicting information.

16. Association Is Not the Same as Causation

Regulatory documents may use carefully calibrated terminology.

Examples include:

These terms should not automatically be converted into:

"Causality has been proven."

The degree of certainty should be preserved.

A good pharmacovigilance summary should reproduce the level of certainty expressed by the evidence.

17. Known, Supported, Plausible and Uncertain

A useful analytical vocabulary is:

Established

Supported sufficiently for the conclusion to be accepted.

Strongly supported

Evidence substantially favours the conclusion, although residual uncertainty remains.

Supported

Evidence favours the conclusion but limitations remain important.

Plausible

The hypothesis is clinically or biologically credible but evidence is insufficient for a strong conclusion.

Uncertain

Available evidence does not permit a reliable determination.

This is especially useful when discussing pharmacological mechanisms.

For example:

The pharmacological effect provides a plausible mechanism for the observed adverse event.

is more scientifically defensible than:

The mechanism proves that the medicine causes the event.

18. Exposure Is Essential

Risk interpretation requires consideration of exposure.

Look for:

The same number of adverse-event reports can represent very different levels of concern depending on exposure.

Where exposure estimates are presented, examine the assumptions used to generate them.

Exposure estimates themselves may carry uncertainty.

19. Relative and Absolute Risk

Regulatory assessments may report:

These measures answer different questions.

A large relative increase can correspond to a small absolute excess.

Conversely, a modest relative increase may have considerable public-health implications when a medicine is widely used.

A complete reading therefore asks both:

How much does the risk increase?

and:

How many additional patients may be affected?

20. Dose and Duration

Dose and duration can affect both causal interpretation and risk management.

Look for:

A credible exposure-response relationship may be particularly influential because it can provide a basis for practical risk reduction.

If therapeutic benefit remains meaningful at lower exposure, dose restriction may become a proportionate option.

21. Risk Factors and Vulnerable Populations

A referral may identify groups at increased risk, such as:

Subgroup findings should not automatically be treated as definitive.

Ask:

22. The Role of Mechanistic Evidence

Pharmacological plausibility can be highly informative.

For example, if an active substance:

the mechanism may make a clinical observation more plausible.

But three different statements should be kept separate:

  1. The drug has a pharmacological effect.
  2. That effect provides a plausible mechanism for the adverse event.
  3. The clinical evidence demonstrates that the mechanism produces a clinically important risk.

The first does not automatically establish the second, and the second does not automatically establish the third.

This distinction should be maintained throughout a pharmacovigilance assessment.

23. Historical Example: Domperidone

The EU referral concerning domperidone-containing medicines provides a useful example of how to reconstruct a regulatory safety assessment.

The procedure was an Article 31 referral concerning nationally authorised products.

The review was initiated in 2013 at the request of the Belgian medicines authority because of concerns about cardiac effects.

The public EMA record contains the notification, review-start documentation, list of questions, PRAC assessment report, PRAC recommendation, CMDh documents and the European Commission decision.

The procedure therefore provides an unusually useful example of the complete evidence-to-action pathway.

24. What Was the Safety Question?

The concern involved cardiac effects including:

The regulatory question was broader than simply asking whether domperidone could affect cardiac repolarisation.

The clinically important questions included:

This illustrates an important principle:

A regulatory safety assessment often progresses from hazard identification to risk characterisation and then to risk management.

25. Reading the Domperidone Evidence

The evidence can be considered by category.

Clinical cases

Serious cardiac events provided important pharmacovigilance information.

Pharmacology

Electrophysiological evidence supported biological plausibility.

Epidemiology

Comparative data contributed information on the magnitude of risk.

Risk factors

Evidence concerning higher-risk circumstances helped determine whether risk could be reduced by restricting exposure.

Benefit-risk

The final regulatory question included whether continued use under restricted conditions remained justified.

The significance lies in the integration of these evidence streams.

26. What the Domperidone Review Demonstrates

The EMA's public summary describes a small increased risk of serious cardiac adverse reactions and identifies higher-risk circumstances, including older age, higher oral doses and concomitant use of QT-prolonging medicines or CYP3A4 inhibitors.

The review considered multiple evidence sources, including non-clinical and clinical data, a thorough QT study, cumulative case-report review, safety-database analyses, pharmacoepidemiological studies and efficacy studies.

The regulatory conclusion was not simply that domperidone should disappear from clinical use.

The CMDh concluded by majority that the medicines should remain available for nausea and vomiting but with restrictions on indication, dose, duration and use in certain patients and with certain concomitant medicines.

Because the CMDh position was adopted by majority vote, it was sent to the European Commission, which issued the final legally binding decision.

This is an excellent example of why a referral should be read as an integrated benefit-risk assessment rather than as a binary safety verdict.

27. From Risk to Restriction

The regulatory response illustrates an important principle.

Regulatory action does not necessarily mean that a medicine must be withdrawn because a serious risk exists.

A medicine may remain available if:

The regulatory response should therefore be understood as a benefit-risk decision rather than simply a declaration of safety or danger.

28. Reading Product Information After the Assessment

After understanding the scientific assessment, examine the product-information changes.

Look for changes to:

Then ask:

Which evidence justified each change?

This creates a direct evidence-to-action chain:

Evidence
   |
   v
Risk characterisation
   |
   v
Risk factors / exposure conditions
   |
   v
Regulatory assessment
   |
   v
Risk-management measure
   |
   v
Product information

This is one of the most useful ways to read a referral from a pharmacovigilance perspective.

29. Historical Example: Valproate

The EU referral concerning valproate and related substances provides a different type of pharmacovigilance lesson.

The major risks associated with valproate exposure during pregnancy were already established.

The regulatory question therefore was not simply:

Does valproate cause fetal harm?

The more important questions included:

The 2017 Article 31 referral was initiated at the request of the French medicines regulator. PRAC conducted the review and issued recommendations, which were considered by CMDh. The CMDh position was subsequently followed by an EU-wide legally binding European Commission decision.

This is a useful example of a referral in which the central regulatory problem involved not merely identification of a serious risk but the prevention of avoidable exposure.

30. Risk Identification Versus Risk Control

The distinction can be expressed as follows:

Question Principal issue
Does the medicine cause the harm? Risk identification
How large is the risk? Risk characterisation
Who is most vulnerable? Risk stratification
Can the harm be prevented? Risk control
Are existing measures working? Risk-minimisation effectiveness
What should change? Regulatory decision

This distinction is particularly important for referrals involving established risks.

A referral may occur because the risk is already known but is not being adequately controlled.

31. Existing Risk Minimisation

When a referral concerns an established risk, ask:

What measures were already in place?

Then ask:

What evidence demonstrates whether those measures were effective?

Relevant evidence may include:

A risk-minimisation measure existing in the product information does not necessarily mean that it is effective in clinical practice.

32. A Referral Question Can Evolve

The initial question and final question may not be identical.

For example:

Initial concern
      |
      v
Is there a safety association?
      |
      v
How large is the risk?
      |
      v
Who is at risk?
      |
      v
Are current measures adequate?
      |
      v
What regulatory action is proportionate?

This evolution is normal.

The evidence collected during the procedure can change the focus of the assessment.

Comparing the initial notification with the final recommendation can therefore reveal how the regulatory reasoning developed.

33. PRAC Recommendation Versus Final Regulatory Outcome

A PRAC recommendation should be identified as such.

It should not automatically be described as the final legally binding outcome.

Depending on the procedure, the subsequent pathway may involve CHMP or CMDh and, where applicable, a European Commission decision.

The EMA's Article 31 pharmacovigilance guidance explains that CHMP or CMDh considers the PRAC recommendation and assessment report and adopts an opinion or position on the maintenance, variation, suspension or revocation of the relevant marketing authorisations.

Where a CMDh position is adopted by majority rather than consensus, it proceeds to the European Commission for a legally binding EU decision.

Therefore, when documenting a historical referral, use precise language:

This avoids conflating scientific recommendations with final legal decisions.

34. CHMP

CHMP is responsible for scientific opinions concerning centrally authorised medicinal products within its remit.

Where the relevant referral pathway requires CHMP involvement, the PRAC recommendation feeds into the CHMP process.

The committees therefore have distinct roles.

PRAC is the specialist committee for pharmacovigilance risk assessment.

CHMP has broader responsibility for scientific opinions concerning medicinal products within its remit.

35. CMDh

CMDh is relevant to nationally authorised medicines within the mutual-recognition and decentralised regulatory framework.

Where applicable, CMDh considers the PRAC recommendation and adopts the appropriate position.

The reader should therefore determine the authorisation route before assuming which committee provides the next regulatory step.

Historical referrals can be particularly instructive because the procedural documents may show whether the CMDh position was reached by consensus or majority and whether divergent positions were recorded.

36. The European Commission Decision

Where a binding EU decision is required, the European Commission adopts the final decision.

The decision may establish measures such as:

The Commission decision should be read separately from the scientific assessment because its function is different.

37. Why the Press Release Is Not Enough

A public summary may explain the conclusion in a few paragraphs.

It may not explain:

For professional pharmacovigilance analysis, the assessment report is therefore usually more informative than the press release.

The press release is the orientation document.

The assessment report is where the detailed reasoning should be examined.

38. A Four-Document Rapid Review

When time is limited, start with four documents:

1. Notification

What triggered the procedure?

2. List of questions

What needed to be answered?

3. Assessment report

How was the evidence interpreted?

4. Final decision

What changed?

This four-document approach usually provides the essential regulatory narrative before more detailed document review.

39. A Six-Question Framework

A useful summary framework is:

1. What happened?

Define the safety concern.

2. What was new?

Identify the evidence that changed the regulatory situation.

3. How strong was the evidence?

Evaluate clinical, epidemiological, pharmacological and other evidence.

4. What remained uncertain?

Identify limitations and contradictory evidence.

5. What did the regulator decide?

State the scientific recommendation and final regulatory outcome separately.

6. Why was the action proportionate?

Relate the action to risk, benefit, preventability and available alternatives.

If these six questions can be answered, the referral has usually been understood at a useful professional level.

40. Regulatory Decisions Are Not Simple Statistical Tests

A regulatory decision is not determined by whether a single p-value crosses a threshold.

A committee may consider:

A statistically uncertain result may still be important when:

Conversely, statistical significance does not automatically establish a clinically important causal relationship if bias or confounding is likely.

41. Benefit-Risk Reasoning

The relevant regulatory question is rarely:

Is there any risk?

Most effective medicines have risks.

The important question is:

Does the overall benefit-risk balance remain favourable under the proposed conditions of use?

This requires consideration of:

The regulatory measure should therefore be interpreted in this context.

42. Why Restriction May Be Preferred to Withdrawal

Suppose a medicine has:

Restriction may preserve therapeutic benefit while reducing exposure to higher-risk circumstances.

Conversely, if:

more severe regulatory action may be considered.

This is why the regulatory outcome should be interpreted as a proportionate response to the complete benefit-risk picture.

43. Reading the Final Product Information

The final product information is where the regulatory conclusion becomes operational.

Consider what each type of change means.

Contraindication

Use under the specified circumstances is considered unacceptable.

Warning

The risk requires heightened clinical awareness but does not necessarily prohibit use.

Dose restriction

Risk is sufficiently related to exposure that dose modification can reduce it.

Duration restriction

Longer exposure contributes materially to risk.

Monitoring

Early detection or intervention may reduce consequences.

Additional risk minimisation

Routine product information alone is considered insufficient to manage the risk.

Reading the product information in this way helps connect the regulatory wording to the underlying evidence.

44. Reading a Referral Backwards

A useful technique is to start with the final outcome.

Ask:

What changed?

Then:

What risk did the change address?

Then:

What evidence demonstrated that risk?

Then:

What regulatory question required that evidence?

Finally:

What triggered the procedure?

Afterwards, verify the reconstruction by reading the documents chronologically.

This combination of backward and chronological reading is particularly effective for complex historical referrals.

45. Historical Reconstruction Without Hindsight Bias

Historical regulatory analysis should respect the information available at the time.

Do not write:

The medicine was known to cause the adverse event, so the regulator acted.

Instead reconstruct:

At the beginning of the procedure, what evidence was available?

Then:

What additional evidence was considered?

Then:

How did the assessment change the degree of certainty?

Finally:

How did the final evidence support the regulatory action?

This makes historical pharmacovigilance analysis scientifically more rigorous.

46. Evidence Evolution

A useful historical framework is:

Early stage

Intermediate stage

Mature stage

The final understanding of a risk should not be projected backwards into the earliest stage.

47. Conflicting Evidence

When studies disagree, do not resolve the disagreement by counting positive and negative studies.

Instead examine:

Two apparently conflicting studies may be answering different questions.

Alternatively, one may have materially stronger internal validity.

The assessment report should be read to determine how these differences were handled.

48. Hazard Versus Risk

A medicine may possess a hazard without creating the same level of clinical risk under every exposure condition.

Hazard

The capacity to cause harm.

Risk

The probability and consequences of harm under specified conditions.

A referral may therefore move from:

Is there a hazard?

to:

Under what circumstances does the hazard translate into clinically important risk?

That transition is central to regulatory pharmacovigilance.

49. Risk Versus Risk-Minimisation Failure

A known risk can be considered manageable if appropriate controls work.

The regulatory problem changes when the controls fail.

Known risk
    |
    v
Existing risk minimisation
    |
    v
Is exposure adequately controlled?
    |
    +---- Yes ----> Continue / monitor
    |
    +---- No -----> Reassess measures
                         |
                         v
                    Additional action

This distinction is particularly important in pregnancy prevention, monitoring programmes, contraindications and restricted-use conditions.

50. Building an Evidence-to-Action Map

For a complex referral, construct a table such as:

Evidence Interpretation Regulatory implication Possible action
Serious clinical cases Consistent adverse-event phenotype Potential clinically important risk Further assessment
Epidemiological association Increased comparative risk Risk may be quantifiable Risk characterisation
Dose-response Higher exposure associated with greater risk Exposure modification may reduce risk Dose restriction
Defined risk factors Risk concentrated in particular patients Targeted mitigation may be feasible Contraindication or warning
Risk-minimisation effectiveness data Existing measures insufficient Current controls inadequate Additional risk minimisation

The actual regulatory decision must always be taken from the individual procedure.

The table is an analytical teaching tool.

51. What Makes a Regulatory Assessment Persuasive?

The strongest safety assessments commonly contain several forms of convergent evidence.

These may include:

Not every safety concern will contain all of these elements.

The important principle is:

Regulatory confidence usually comes from the totality of evidence rather than a single observation.

52. What Makes a Referral Difficult?

The most difficult referrals often contain:

These are precisely the cases in which regulatory reasoning is most instructive.

The committee must make a decision despite residual uncertainty.

The reader should therefore study not only the conclusion but the way uncertainty was managed.

53. Historical Referrals as Teaching Cases

Historical referrals provide a particularly useful method for teaching pharmacovigilance.

The final outcome is known.

The evidence at each stage can nevertheless be reconstructed.

This permits the reader to ask:

Would the available evidence at that time have justified the same conclusion?

That question helps develop judgement.

It also highlights common analytical errors:

54. Two-Pass Reading Method

A two-pass approach is particularly effective.

Pass 1: Scientific assessment

Read for:

Where possible, form your own provisional assessment before reading the final regulatory decision.

Pass 2: Regulatory assessment

Then assess:

Finally compare your provisional scientific interpretation with the regulatory conclusion.

This makes the referral an active learning exercise rather than passive reading.

55. Using EMA Referrals in Signal Evaluation

An EMA referral can be important external evidence when evaluating a related signal.

However, the conclusion should not simply be copied.

Assess whether:

A historical regulatory conclusion provides important context.

It does not eliminate the need for product-specific assessment.

56. Using Referrals in PBRERs and PSURs

A relevant referral may affect periodic safety evaluation.

Consider whether it changes:

The referral should therefore be integrated into the lifecycle safety assessment.

57. Using Referrals in RMPs

A referral may result in changes to:

The RMP should reflect the current regulatory state rather than the state before the referral.

58. What a QPPV Should Ask

When an important referral is identified, the QPPV should be able to answer:

  1. What triggered the procedure?
  2. Is the issue relevant to our products?
  3. What evidence was most influential?
  4. What evidence remained uncertain?
  5. Which patients are most affected?
  6. Has the benefit-risk balance changed?
  7. Why was the regulatory measure chosen?
  8. Do our products have the same relevant risk factors?
  9. Is our product information aligned?
  10. Does our RMP remain appropriate?
  11. Are risk-minimisation measures effective?
  12. Does signal management need to change?

These questions convert regulatory intelligence into pharmacovigilance oversight.

59. Inspection Perspective

An inspector may ask:

How does your organisation evaluate regulatory safety information such as EMA referrals?

A mature system should be able to demonstrate:

The evidence should demonstrate not merely that the organisation receives EMA information, but that it interprets and acts on information relevant to its products.

60. Common Reading Errors

60.1 Reading only the press release

This removes much of the scientific reasoning.

60.2 Reading only the final decision

This shows what changed but not why.

60.3 Treating referral initiation as proof of causality

A referral establishes a formal assessment process, not a causal conclusion.

60.4 Treating all evidence types as equivalent

A case report, randomised trial and epidemiological study answer different questions.

60.5 Ignoring negative evidence

Contradictory evidence is essential for understanding uncertainty.

60.6 Ignoring exposure

Report counts without denominator context can mislead.

60.7 Ignoring absolute risk

Relative measures alone can distort clinical interpretation.

60.8 Ignoring benefit

Regulatory decisions concern benefit-risk, not hazard in isolation.

60.9 Assuming a known risk requires withdrawal

Risk may be manageable through appropriate restrictions.

60.10 Using hindsight

The evidence available at the end of the procedure should not be assumed to have been available at the beginning.

The subsequent regulatory pathway must be followed.

60.12 Assuming a referral automatically applies to another product

The scope of the procedure and relevance to the individual product must be assessed.

61. A Practical Referral Reading Checklist

Procedure

Trigger

Scientific question

Evidence

Risk

Uncertainty

Regulatory conclusion

Implementation

MAH relevance

62. Final Perspective

The most useful way to read an EMA referral is as a chain of reasoning:

trigger → regulatory question → evidence → interpretation → uncertainty → benefit-risk → recommendation → decision → implementation

The notification explains why the process began.

The questions show what regulators wanted to know.

The assessment report explains how the evidence was evaluated.

The PRAC recommendation records the pharmacovigilance conclusion.

The subsequent regulatory stages establish the applicable regulatory outcome.

The product-information changes show how the conclusion was translated into clinical use.

This approach also explains why an EMA referral should not be treated simply as evidence that:

"The medicine causes the event."

The more precise question is:

What did the regulator conclude, on what evidence, with what degree of uncertainty, and what action was considered proportionate?

That is the level at which a pharmacovigilance professional should read regulatory safety assessments.

Historical referrals are particularly valuable because they allow the evidence to be reconstructed over time.

The reader can distinguish:

In that sense, an EMA referral is not merely a regulatory document.

It is a case study in applied pharmacovigilance reasoning.

Key Takeaways

References

  1. European Medicines Agency. Questions and answers: Article 31 pharmacovigilance referrals. EMA.

  2. European Medicines Agency. Good pharmacovigilance practices (GVP), Module IX – Signal management. EMA.

  3. European Medicines Agency. Good pharmacovigilance practices (GVP), Module V – Risk management systems. EMA.

  4. European Medicines Agency. Good pharmacovigilance practices (GVP), Module I – Pharmacovigilance systems and their quality systems. EMA.

  5. European Parliament and Council. Directive 2001/83/EC on the Community code relating to medicinal products for human use, as amended.

  6. European Parliament and Council. Regulation (EC) No 726/2004, as amended.

  7. European Commission. Commission Implementing Regulation (EU) No 520/2012 on the performance of pharmacovigilance activities, as amended.

  8. European Medicines Agency. Domperidone-containing medicines – referral. Article 31 referral, EMEA/H/A-31/1365.

  9. European Medicines Agency. Domperidone Article-31 referral – PRAC assessment report. EMA/152501/2014.

  10. European Medicines Agency. Domperidone Article-31 referral – Restrictions on the use of domperidone-containing medicines. EMA/465179/2014.

  11. European Medicines Agency. Domperidone Article-31 referral – PRAC recommends restricting use of domperidone. EMA/129231/2014.

  12. European Medicines Agency. Domperidone Article-31 referral – CMDh documents and divergent position. EMA.

  13. European Medicines Agency. Valproate and related substances – referral. Article 31 referral, EMEA/H/A-31/1454.

  14. European Medicines Agency. Valproate Article-31 referral – PRAC recommends new measures to avoid valproate exposure in pregnancy. EMA/67672/2018.

  15. European Medicines Agency. Valproate Article-31 referral – PRAC assessment report. EMA/198940/2018.

  16. European Medicines Agency. Valproate Article-31 referral – New measures to avoid valproate exposure in pregnancy endorsed. EMA/145600/2018.

  17. Council for International Organizations of Medical Sciences (CIOMS). Practical Aspects of Signal Detection in Pharmacovigilance: Report of CIOMS Working Group VIII. CIOMS.

Regulatory interpretation note

This article describes how to read and analyse publicly available EMA referral documentation. It is an educational resource and does not replace applicable legislation, European Commission decisions, EMA procedural guidance, GVP requirements or product-specific regulatory advice.

The evidence-to-action maps, reading frameworks and checklists in this article are teaching tools developed to structure pharmacovigilance analysis. They are not EMA-mandated assessment templates or regulatory scoring systems.

Historical examples are used to illustrate regulatory reasoning. The evidence available during the historical procedure should be distinguished from information that became available subsequently.

Where the procedural pathway or legal consequence depends on the individual referral, the applicable EMA documentation and legislation should be consulted directly.

Revision History

Related Articles

How PRAC Assesses a Safety Issue During an EMA Referral

How the Pharmacovigilance Risk Assessment Committee evaluates a referred safety issue, including the assessment framework, rapporteur and co-rapporteur roles, evidence sources, benefit-risk reasoning and development of the PRAC recommendation.

Historical Signal Evaluation: Metamizole and Agranulocytosis

A reconstruction of the 2024 EU metamizole–agranulocytosis review, from the regulatory trigger and available evidence through PRAC assessment, risk-minimisation conclusions and changes to product information.

What Is an EMA Referral? Why Referrals Are Started and How the EU Referral System Works

A practical and regulatory guide to EMA referral procedures, distinguishing safety signals from formal referrals and explaining the roles of PRAC, CHMP, CMDh and the European Commission.