QPPV.COM

Emerging Safety Issue Assessment: Evidence, Urgency and Regulatory Decision-Making

How to evaluate a potential emerging safety issue using structured evidence assessment, clinical judgement, causality, magnitude, urgency and regulatory context, with practical examples and documented reasoning.

Audio Lesson 17 min
Knowledge Assessment Test your understanding of this article. Take the assessment →

Emerging Safety Issue Assessment: Evidence, Urgency and Regulatory Decision-Making

Introduction

An emerging safety issue (ESI) is not simply a serious adverse reaction and it is not synonymous with a validated safety signal. It represents a safety concern that may require attention on a timescale shorter than the ordinary pharmacovigilance process because of its potential major impact on the benefit-risk balance of a medicinal product and/or on patients or public health.

This article addresses the difficult part of the process: how to evaluate whether a safety concern warrants ESI treatment.

The distinction matters. A pharmacovigilance system may identify hundreds of signals over time, while only a small proportion will require urgent regulatory attention. Conversely, a potentially important safety concern may arise from a clinical study, epidemiological analysis, scientific publication or regulatory action before a conventional signal-management process has identified it.

The assessment therefore cannot be reduced to a statistical threshold or a fixed number of cases.

It requires integration of:

The objective is not to establish certainty before action.

The objective is to make a defensible judgement using the information available at the time.

This article complements Emerging Safety Issues in Pharmacovigilance, which addresses the operational workflow, governance, escalation, documentation and inspection aspects of ESI management. The present article concentrates on the scientific and regulatory reasoning behind the ESI decision.

Learning Objectives

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

1. Regulatory Definition

1.1 The ESI concept

GVP Module IX describes an emerging safety issue as a safety issue considered by the marketing authorisation holder (MAH) to require urgent attention by the competent authority because of its potential major impact on the risk-benefit balance of the medicinal product and/or on patients' or public health, and the potential need for prompt regulatory action and communication to patients and healthcare professionals.[1]

Several elements of the definition are important.

First, there must be a safety issue. The concept is not intended to function as a general escalation mechanism for every operational problem.

Second, the issue must have potential major impact. The concern therefore has to be considered in relation to its possible clinical and public-health consequences.

Third, the issue must require urgent attention. This is the element that distinguishes an ESI from many ordinary signals.

Fourth, the definition refers to the potential need for regulatory action and communication. A final regulatory conclusion does not have to exist before an issue can warrant urgent attention.

The definition therefore describes a decision under uncertainty.

2. What an ESI Is Not

An ESI should not be treated as a synonym for:

These concepts may overlap, but they answer different questions.

A serious adverse reaction concerns the characteristics of an individual adverse reaction.

A signal concerns information suggesting a new potentially causal association or a new aspect of a known association.

Signal validation asks whether there is sufficient evidence to justify further analysis.

Signal assessment asks what the totality of evidence indicates.

An ESI asks whether the safety issue is sufficiently important and urgent that competent-authority attention should not wait for the ordinary timetable.

3. Signal Evaluation

3.1 Signal detection is not ESI assessment

Signal detection is designed to identify potential safety concerns.

Methods may include:

Detection is deliberately sensitive.

ESI assessment is different. It requires a judgement about:

A signal can therefore be statistically interesting without being an ESI.

An ESI can also arise without a conventional statistical signal.

3.2 Signal validation

Signal validation asks whether the available evidence supports further analysis of a potentially causal association or a new aspect of a known association.[1]

Validation may consider:

A validated signal is not automatically an ESI.

The next question is whether the concern has characteristics that make urgent regulatory attention appropriate.

3.3 Signal prioritisation

Signal prioritisation considers which validated signals require further action and with what urgency.

This is closer to ESI assessment than signal detection is, but the concepts remain distinct.

A signal may be prioritised highly because it is:

It may still be possible to conduct the assessment through the ordinary signal-management pathway.

The ESI threshold is reached when the urgency and potential consequences make delay particularly important.

4. The Central Question: Can We Safely Wait?

A useful practical formulation is:

Could patients or public health be exposed to avoidable serious harm if the issue were handled only through the ordinary safety-assessment timetable?

This is not a substitute for the regulatory definition. It is a way of making its urgency component operational.

The question should be followed by another:

Would earlier regulatory awareness plausibly change the opportunity to prevent or reduce harm?

If the answer is yes, the issue deserves particularly careful ESI consideration.

This approach avoids two common errors.

The first is excessive caution:

"The evidence is not yet conclusive, so nothing can be done."

The second is excessive alarm:

"The event is serious, so it must be an ESI."

Neither is adequate.

5. Evidence Available at the Time

5.1 Temporal discipline

ESI assessment should be based on the information available when the decision is made.

This becomes particularly important when evaluating historical cases.

A later regulatory conclusion may provide useful context, but it should not be used to imply that the same conclusion was obvious at the beginning.

The assessment should therefore distinguish:

T0

What was known when the concern was first identified?

T1

What additional evidence became available during assessment?

T2

What regulatory interpretation followed?

T3

What did later evidence establish?

This prevents hindsight bias.

5.2 A historical evidence table

Time point Evidence available Appropriate interpretation
T0 Initial case, study or analysis What could reasonably be suspected
T1 Additional cases or analyses Whether the concern strengthened or weakened
T2 Regulatory assessment How authorities interpreted the evidence
T3 Subsequent evidence Whether the original concern was confirmed, narrowed or refuted

A historical article should never collapse these stages into a single retrospective conclusion.

6. Start With What Changed

A good assessment begins with a precise description of the new information.

Weak:

"There is a potential safety signal."

Better:

"A newly completed study identified an unexpected imbalance in fatal hepatic events between treatment and control groups."

Or:

"A cluster of clinically similar cases has been identified in a previously unrecognised patient population."

Or:

"A regulatory authority outside the EU has introduced a new contraindication following review of serious cardiac events."

The first task is description.

Interpretation comes next.

7. Clinical Characterisation

7.1 Seriousness and severity

The clinical consequence of the event should be described precisely.

Relevant features include:

"Serious" and "severe" are not interchangeable terms.

Seriousness is a regulatory concept.

Severity describes the intensity of an event.

The assessment should use the appropriate terminology rather than treating the words as synonyms.

7.2 Reversibility

Reversibility can materially affect urgency.

A transient laboratory abnormality that is detected early may have a different clinical significance from an injury that becomes irreversible before it is recognised.

Questions include:

The consequence of delayed recognition should be considered explicitly.

8. Time to Onset

Latency is often one of the most informative clinical features.

A plausible temporal relationship may support an association.

An implausible temporal relationship may weaken it.

The assessment should consider:

A coherent latency pattern does not prove causality.

It does, however, contribute to the overall evidence.

9. Dechallenge and Rechallenge

Where relevant, the assessment should consider:

Dechallenge

Did the event improve after treatment was stopped?

Rechallenge

Did the event recur after treatment was restarted?

A positive rechallenge can provide strong evidence in some clinical circumstances.

However, rechallenge should never be treated as a requirement for signal validity or ESI classification.

Intentional rechallenge may be inappropriate or unethical where the suspected reaction is severe.

The absence of rechallenge evidence therefore has limited negative value in many serious reactions.

10. Alternative Explanations

A credible ESI assessment actively considers competing explanations.

Possible alternatives include:

The question is not whether every alternative explanation can be excluded.

The question is whether the alternatives are sufficiently credible to alter the urgency of the concern.

11. Case-Level Assessment

When the concern arises from spontaneous reports, the clinical quality of the individual cases matters.

Useful features include:

A large number of poorly characterised cases can be less informative than a smaller number of clinically coherent cases.

Conversely, incomplete spontaneous reports should not automatically be dismissed where the potential consequence is substantial.

12. Case Series

Case series become more informative when they show:

The reviewer should also ask whether cases were identified through a mechanism that could create reporting enrichment.

For example, a widely publicised adverse event may produce a sudden increase in reporting that reflects awareness rather than an increase in incidence.

The signal therefore requires clinical interpretation, not simply counting.

13. Exposure and Denominators

Case counts have little meaning without exposure context.

The assessment should consider:

A cluster of 20 events can have very different implications in:

Similarly, a rare event may be highly important if the outcome is catastrophic and the exposed population is large.

14. Vulnerable Populations

Particular attention may be warranted when the concern affects:

A small number of cases in a vulnerable population may justify greater concern than a larger number in a lower-risk population.

The assessment should therefore consider both the absolute number of patients and the characteristics of those patients.

15. Biological and Pharmacological Plausibility

Pharmacological plausibility can strengthen a safety hypothesis.

Relevant evidence may include:

But plausibility must not be confused with proof.

A useful distinction is:

Established

The biological property is demonstrated.

Plausible

The property could reasonably explain the clinical observation.

Supported

Clinical and experimental evidence converge on the proposed mechanism.

Uncertain

The mechanism remains incompletely understood.

Unsupported

The proposed mechanism lacks credible evidence.

This distinction should be maintained throughout the assessment.

16. What the Mechanism Can and Cannot Tell Us

Suppose a medicine has a pharmacological effect that could theoretically produce a particular organ injury.

That supports plausibility.

It does not establish:

Mechanistic reasoning is therefore most useful when integrated with clinical and epidemiological evidence.

17. Quantitative Evidence

17.1 Disproportionality analysis

Disproportionality methods can identify reporting patterns that merit investigation.

They can be useful for:

They do not directly provide:

A high disproportionality measure can therefore identify a signal without establishing that the issue is an ESI.

17.2 The denominator problem

Spontaneous reporting systems generally do not provide a complete denominator of exposed patients.

Consequently:

number of reports ≠ incidence

and:

disproportionality ≠ relative risk.

This distinction is fundamental.

An ESI assessment should avoid converting spontaneous-report metrics into unsupported estimates of population risk.

18. Epidemiological Evidence

Epidemiological studies can provide information that spontaneous reporting cannot.

They may estimate:

The assessment should consider:

A modest relative risk can still have substantial public-health implications when exposure is widespread and the outcome is severe.

19. Clinical-Trial Evidence

Clinical trials provide structured exposure and follow-up, but their limitations must also be considered.

A trial may be particularly informative where:

However, rare events may remain difficult to interpret when:

A serious unexpected imbalance in a large randomised study may nevertheless require urgent attention before every uncertainty has been resolved.

20. Literature Evidence

Scientific literature can generate or strengthen an emerging safety concern.

Potentially important publications include:

The assessment should distinguish the quality of the publication from its prominence.

A highly cited paper is not necessarily methodologically strong.

A single well-designed study may be more informative than many low-quality case reports.

21. Regulatory Information

A regulatory action in another jurisdiction can itself constitute important safety information.

The MAH should determine:

The existence of an overseas regulatory action does not automatically establish the same conclusion for the EU.

It does, however, require prompt assessment.

22. Known Risk Versus New Risk

A new report of a known adverse reaction is not necessarily a new signal.

A concern may instead involve a new aspect of a known association.

Examples include evidence that:

This distinction is particularly important when evaluating whether a change in the safety profile could materially affect benefit-risk.

23. Unexpectedness

Unexpectedness should be assessed against the current safety information.

Relevant sources include:

An event may be clinically serious but not unexpected.

Conversely, a previously unrecognised event may be unexpected even when only limited cases exist.

Unexpectedness alone does not establish an ESI.

24. Potential Magnitude

Potential magnitude includes more than observed frequency.

Consider:

A rare catastrophic event associated with a medicine used by millions of people may warrant greater urgency than a more frequent event affecting a very small population.

The potential number of future affected patients can therefore matter even when current case counts are small.

25. Public-Health Impact

Some concerns have implications beyond individual patients.

Examples include:

Public-health impact should be considered independently from the number of currently reported cases.

26. Preventability

The potential to reduce harm can materially affect urgency.

Possible interventions include:

If a serious risk is both credible and readily preventable, earlier regulatory attention may have greater potential value.

27. Detectability and Opportunity for Intervention

A risk is more concerning when serious harm occurs before clinicians have an opportunity to recognise and intervene.

Questions include:

This helps distinguish severity from urgency.

28. Benefit-Risk Context

A safety concern should never be assessed in isolation from the medicine's benefits.

Relevant considerations include:

The same adverse event may have different implications in different therapeutic settings.

A rare fatal risk associated with a treatment for a life-threatening disease may produce a different regulatory response from the same risk associated with treatment of a mild condition.

29. Regulatory Action Does Not Mean Withdrawal

Potential regulatory responses are broader than suspension or withdrawal.

They may include:

The ESI decision therefore asks whether urgent regulatory attention is warranted, not whether the product should necessarily be removed from the market.

30. Consequences of Waiting

The urgency assessment should explicitly consider the counterfactual.

Ask:

What could happen if the issue remains within the routine safety process for several weeks?

Then consider:

This is one of the most useful ways to distinguish a clinically important signal from an ESI.

31. Causality and Urgency Are Different Dimensions

Causal certainty and urgency should be considered separately.

Causal evidence Potential impact Possible interpretation
Weak Low Routine monitoring may be appropriate
Strong Low Important signal, but not necessarily urgent
Weak/moderate High Urgent attention may be warranted
Strong High Strong basis for urgent regulatory consideration

This is not a regulatory scoring system.

It illustrates a fundamental principle:

Low causal certainty does not necessarily mean low urgency.

If potential harm is severe, preventable and widespread, the consequences of waiting may justify early regulatory attention while the evidence continues to develop.

32. Negative Evidence

A rigorous assessment should actively search for evidence against the safety hypothesis.

This may include:

Negative evidence should not be treated as an afterthought.

The purpose of assessment is to determine what the totality of evidence supports.

33. Competing Hypotheses

A useful assessment may explicitly compare competing explanations.

For example:

Hypothesis Supporting evidence Evidence against
Treatment causes the event Compatible latency; repeated phenotype Limited case numbers
Underlying disease causes the event Disease is independently associated Event pattern differs from expected background
Concomitant medicine causes the event Known association Not present in all cases
Reporting artefact Publicity increased reporting Objective study finding also present

The objective is not to produce a mathematical probability.

It is to make the reasoning transparent.

34. When a Signal Is Not an ESI

Consider a validated signal involving a non-serious adverse reaction.

Suppose:

This can be a legitimate signal requiring assessment without being an ESI.

A mature pharmacovigilance system must be able to reach this conclusion without treating non-ESI classification as dismissal of the signal.

35. When an ESI May Arise Before Signal Validation

The reverse is also possible.

Suppose a newly completed randomised trial identifies an unexpected excess of fatal events.

The issue may be identified through clinical-development safety surveillance rather than conventional post-marketing signal detection.

There is no scientific reason to wait for:

If the information itself potentially meets the ESI definition, it should be assessed as such.

36. Historical Example: Valproate

Valproate provides an instructive example of how an established risk can evolve into a major regulatory problem when evidence shows that existing risk-minimisation measures are insufficient.

EMA's 2014 review strengthened restrictions because of the recognised risks of congenital malformations and developmental problems following exposure during pregnancy.[2]

A subsequent EU review, initiated in 2017, considered whether existing measures were sufficiently effective. PRAC concluded in 2018 that further measures were necessary because women were still not always receiving the appropriate information in a timely manner and unintended exposure continued to occur.[3][4]

The regulatory response included stronger restrictions and a pregnancy prevention programme.

The important teaching point is not simply that valproate is teratogenic.

That risk was already established.

The more sophisticated question is:

What changed in the safety-management problem when evidence indicated that existing measures were not adequately preventing exposure?

This illustrates a critical distinction between:

The latter can materially alter benefit-risk management even when the underlying adverse effect is already known.

37. Historical Example: Domperidone

The EU review of domperidone considered serious cardiac adverse reactions, including QT prolongation, ventricular arrhythmia and sudden cardiac death.[5]

The assessment identified circumstances associated with increased risk, including higher doses, older age and concomitant medicines that could prolong the QT interval or increase domperidone exposure.

The regulatory response involved restrictions on indication, dose and treatment duration, together with contraindications and warnings.

The teaching point is important:

A serious safety finding does not automatically lead to withdrawal.

The regulatory question is whether the risk can be sufficiently reduced through changes in the conditions of use.

This is benefit-risk reasoning rather than simple hazard identification.

38. Historical Example: Fluoroquinolone-Associated Disabling Reactions

The EU review of quinolone- and fluoroquinolone-containing medicines considered serious, disabling and potentially permanent adverse effects.[6]

The resulting regulatory action included restrictions on use and strengthened information concerning these risks.

The case illustrates several dimensions relevant to ESI assessment:

It also demonstrates why the eventual regulatory action should not be treated as proof that the same conclusion was obvious at the earliest stage of the safety concern.

39. Historical Analysis Must Avoid Hindsight Bias

When using historical regulatory cases, distinguish:

What was known at the time

from:

What became known later.

For example, if a later epidemiological study confirms a risk, that study should not be inserted into the evidence base available when the original safety concern was first assessed.

A historically rigorous analysis asks:

What could a competent pharmacovigilance professional reasonably have concluded at that point in time?

This approach is particularly valuable when teaching ESI assessment.

40. Historical Product Information as Evidence

Changes in product information can help reconstruct regulatory evolution.

A useful comparison is:

Stage Question
Earlier product information What risk was recognised?
New evidence What information changed?
Regulatory assessment How was the new evidence interpreted?
Later product information What warning, restriction or contraindication was introduced?

This approach can demonstrate how pharmacovigilance evidence is translated into clinical instructions.

41. Regulatory Language Matters

Regulatory wording should be interpreted carefully.

There is a difference between:

These expressions do not necessarily represent the same level of certainty or regulatory consequence.

A historical analysis should reproduce the meaning of regulatory language accurately rather than treating every change as equivalent.

42. ESI Notification

Under the published GVP Module IX Rev. 1, when an EU MAH becomes aware of an ESI from any source, it should notify the relevant competent authorities and EMA in writing as soon as possible and no later than three working days after establishing that a validated signal or safety issue meets the ESI definition.[1]

The notification should describe:

Further relevant information should be provided as it becomes available.[1]

This requirement should be understood in conjunction with the applicable legislation and current regulatory guidance.

43. ESI Notification Does Not Replace ICSR Reporting

If the ESI concerns an individual suspected adverse reaction, applicable ICSR reporting requirements continue to apply.

The two mechanisms have different purposes.

ICSR reporting

Communicates an individual suspected adverse reaction.

ESI notification

Communicates an urgent broader safety concern requiring competent-authority attention.

One does not replace the other.

44. ESI and Regulatory Signal Management

An ESI should also not automatically be equated with the formal EU signal-management process.

GVP Module IX recognises that an ESI may arise from any source and may require urgent attention outside the ordinary signal-management timetable.[1]

The immediate objective is to ensure that the competent authority is aware of a potentially major and urgent safety issue.

The subsequent regulatory pathway depends on the nature of the concern and the evidence.

45. The Role of the QPPV

The QPPV should have sufficient authority and access to ensure awareness of emerging safety concerns and other information relevant to the benefit-risk evaluation of medicinal products.[7]

This means the QPPV should have practical visibility of information such as:

The QPPV does not have to personally perform every scientific analysis.

The QPPV must, however, be able to exercise meaningful oversight of the pharmacovigilance system and respond when a concern may require regulatory action.

46. External QPPV Arrangements

The same principle applies when the QPPV is external.

The relevant question is not whether the QPPV receives a periodic safety summary.

It is whether the arrangement ensures that the QPPV receives relevant information early enough to recognise and escalate an emerging safety concern.

Potential weaknesses include:

An external appointment can therefore be compliant in form but weak in operation if information does not reach the QPPV in time.

47. ESI and Outsourcing

Outsourcing pharmacovigilance activities does not transfer the MAH's responsibility for its pharmacovigilance system.

Contracts and safety-data-exchange arrangements should provide appropriate escalation mechanisms for:

A vendor may identify the concern first.

The MAH and QPPV remain responsible for ensuring that the concern is appropriately governed and escalated.

48. Quality Defects and Other Regulatory Pathways

Not every urgent safety-related problem belongs in the ESI pathway.

Quality defects, suspected falsified medicines and certain supply-related problems may be governed by separate regulatory procedures.

The assessment should therefore first identify the nature of the problem.

A safety concern may require more than one pathway, but the ESI process should not become a generic substitute for:

The correct pathway should be determined from the applicable regulatory framework.

49. A Structured ESI Assessment

The following framework is useful for documenting the reasoning without introducing an artificial numerical score.

49.1 What is new?

Describe the new information in one or two precise sentences.

49.2 What is the clinical consequence?

Assess:

49.3 How credible is the association?

Assess:

49.4 How large could the problem be?

Assess:

49.5 What does the broader evidence show?

Assess:

49.6 Could benefit-risk be materially affected?

Consider:

49.7 What is the consequence of waiting?

Ask:

49.8 Does the concern meet the ESI definition?

Record the conclusion and rationale.

50. Evidence Matrix

A simple evidence matrix can improve transparency.

Domain Evidence supporting concern Evidence against concern Remaining uncertainty
Clinical phenotype
Temporal relationship
Alternative causes
Dechallenge/rechallenge
Exposure
Epidemiology
Clinical trials
Literature
Biological plausibility
Severity
Preventability
Benefit-risk impact
Consequence of delay

This is an analytical tool, not an EMA scoring system.

Its purpose is to make the reasoning visible.

51. Why Artificial Scoring Can Be Misleading

A numerical score may appear to improve consistency:

Severity = 5
Plausibility = 4
Frequency = 3
Population impact = 4

But the regulatory definition does not establish a validated arithmetic relationship between these dimensions.

For example:

A numerical threshold can therefore create false precision.

Structured qualitative reasoning is generally more defensible unless an organisation has a validated, justified methodology for a specific purpose.

52. Two-Pass Assessment

A two-pass approach is useful for high-consequence decisions.

Pass 1: Scientific and clinical assessment

Determine:

The objective is to describe the evidence accurately.

Pass 2: Regulatory urgency

Determine:

Separating these passes reduces the risk that the desire for rapid action will distort the scientific assessment.

53. Worked Hypothetical: A New Fatal Adverse Event

Consider a fictional medicine, Product A.

A newly completed randomised study identifies:

The correct first conclusion is not:

"Product A causes fatal liver injury."

That statement exceeds the evidence.

Instead:

Established

There is an imbalance in fatal hepatic events.

Supporting evidence

Several cases have compatible laboratory findings.

Alternative explanation

The underlying disease and baseline characteristics require assessment.

Clinical significance

The outcome is fatal.

Potential preventability

If the laboratory pattern is confirmed, earlier detection might provide an opportunity for intervention.

Uncertainty

The causal contribution of treatment and the magnitude of the risk remain uncertain.

ESI question

Could the concern have a major effect on benefit-risk or public health, and could waiting for the ordinary assessment expose patients to avoidable harm?

That is the appropriate decision point.

54. What Should Be Investigated Immediately?

The assessment should seek:

The existence of these additional analyses does not automatically justify delaying an ESI notification if the issue already meets the regulatory definition.

55. Worked Hypothetical: A Large Number of Mild Events

Product B is associated with 500 reports of headache.

Further review shows:

The signal may be important to evaluate.

The number of reports alone does not make it an ESI.

This example illustrates why:

Case count is not an ESI threshold.

56. Worked Hypothetical: A Single Catastrophic Case

Product C has no known association with severe anaphylaxis.

One patient develops:

One case does not establish population-level causality.

But it may warrant urgent attention because:

The correct question is not:

"Is one case enough to prove the risk?"

It is:

"Is the information sufficiently concerning that waiting for additional cases may expose patients to avoidable harm?"

57. What the ESI Decision Does Not Establish

An ESI decision does not necessarily establish:

It establishes that the concern warrants urgent regulatory attention under the applicable framework.

The scientific assessment continues.

58. Documentation of the Decision

A defensible ESI record should allow another qualified reviewer to reconstruct the decision.

At minimum, document:

The record should also distinguish information available at the time of the decision from information obtained later.

59. How to Write the Conclusion

A good conclusion should answer three questions:

What is the evidence?

"The available cases demonstrate a consistent clinical phenotype and temporal relationship, although the contribution of underlying disease remains uncertain."

What does it mean?

"The available evidence supports a potential treatment-associated risk but does not yet establish its incidence or magnitude."

Why is it urgent or not urgent?

"Because the outcome is potentially fatal, the exposed population is substantial and earlier intervention could reduce risk, the issue warrants urgent regulatory attention."

Or, where appropriate:

"Although the association warrants further assessment, the available evidence does not indicate a potential major impact on benefit-risk or public health that would make routine assessment inappropriate."

That is much stronger than:

"Signal is positive; ESI confirmed."

60. Inspection Perspective

Inspectors may ask:

How does the organisation determine whether a safety concern is an ESI?

A strong answer describes the reasoning framework rather than merely naming a person or SOP.

The organisation should be able to demonstrate:

The strongest evidence is usually a contemporaneous example.

61. Common Assessment Failures

Treating seriousness as sufficient

A serious adverse event is not automatically an ESI.

Treating causality as sufficient

A strong causal association is not automatically urgent.

Treating uncertainty as a reason not to act

Incomplete evidence can still require urgent attention.

Treating statistics as clinical judgement

A disproportionality measure does not determine severity or benefit-risk impact.

Ignoring negative evidence

An assessment should actively test competing explanations.

Using hindsight

Later evidence should not be used to rewrite the information available at the time.

Confusing ESI with regulatory outcome

An ESI can result in a range of regulatory responses.

Using arbitrary numerical thresholds

False precision can obscure rather than improve judgement.

62. A Practical Assessment Checklist

Before concluding that a concern meets or does not meet the ESI definition, ask:

What changed?

Clinical significance

Evidence

Population

Benefit-risk

Urgency

Governance

Regulatory pathway

63. Final Perspective

The difficult part of emerging safety issue assessment is not recognising that a safety event is serious.

It is deciding what the available evidence means before the answer is complete.

A competent assessment therefore does not ask only:

"Is this causal?"

It asks:

"What is the evidence?"

"What does it support?"

"What does it not support?"

"What could happen if the concern is real?"

"How many patients could be affected?"

"Can the harm be prevented?"

"What would be gained by immediate regulatory attention?"

"What would be lost by waiting?"

These questions place scientific uncertainty and regulatory urgency in the same analytical framework.

That is the essential skill in ESI assessment.

Key Takeaways

An emerging safety issue is an urgency and potential-impact concept, not simply another name for a serious signal.

A validated signal does not automatically constitute an ESI.

An ESI does not require confirmed causality.

The assessment should integrate clinical evidence, temporal relationships, alternative explanations, exposure, epidemiology, clinical-trial findings, literature, pharmacological plausibility and regulatory information.

Pharmacological mechanisms should be described carefully, distinguishing what is established from what is plausible and what remains uncertain.

Case counts and disproportionality measures should not be converted into unsupported estimates of incidence or clinical importance.

Benefit-risk must be considered in therapeutic context.

Preventability, detectability, reversibility and the consequences of delay can materially affect urgency.

Historical cases should be reconstructed using the evidence available at the time and should not be interpreted retrospectively as though the final regulatory conclusion was obvious from the beginning.

An ESI notification does not replace applicable ICSR reporting requirements.

The QPPV should have sufficient access and authority to recognise and escalate emerging safety concerns.

A strong ESI assessment is not alarmist and it is not complacent. It is explicit about evidence, uncertainty, potential consequences and the reason why immediate regulatory attention is—or is not—justified.

References

  1. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP): Module IX – Signal management (Rev. 1). In particular IX.A.1.1, IX.B and IX.C.2. EMA/827661/2011. Current published Module IX specifies notification of an ESI as soon as possible and no later than three working days after establishing that the definition is met.

  2. European Medicines Agency. PRAC recommends strengthening the restrictions on the use of valproate in women and girls. 2014.

  3. European Medicines Agency. PRAC recommends new measures to avoid valproate exposure in pregnancy. EMA/67672/2018. 9 February 2018.

  4. European Medicines Agency. Valproate and related substances – referral. Article 31 referral; procedure initiated 9 March 2017, PRAC recommendation 8 February 2018, CMDh position 21 March 2018 and European Commission decision 31 May 2018.

  5. European Medicines Agency. Domperidone-containing medicines – referral. EU review concerning cardiac risks and resulting restrictions.

  6. European Medicines Agency. Quinolone- and fluoroquinolone-containing medicinal products – referral. EU review concerning serious, disabling and potentially permanent adverse effects and resulting restrictions.

  7. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP): Module I – Pharmacovigilance systems and their quality systems. In particular provisions concerning QPPV responsibilities, access to information on emerging safety concerns and regulatory action.

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

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

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

  11. European Commission. Commission Implementing Regulation (EU) 2025/1466 of 22 July 2025 amending Commission Implementing Regulation (EU) No 520/2012 on the performance of pharmacovigilance activities.

  12. European Medicines Agency. Good pharmacovigilance practices (GVP). Current EMA GVP overview and information concerning forthcoming revisions following Commission Implementing Regulation (EU) 2025/1466.

Regulatory status note

The detailed ESI definition and three-working-day notification provision cited in this article come from the currently published GVP Module IX Rev. 1.[1]

EMA states that Commission Implementing Regulation (EU) 2025/1466 amended the underlying pharmacovigilance framework and that affected GVP modules will be revised accordingly.[11][12] The article therefore distinguishes the currently published GVP text from the evolving legal and guidance framework.

The analytical frameworks in this article—including the evidence matrix, two-pass assessment and counterfactual questions—are practical teaching tools. They are not presented as EMA-mandated scoring systems or regulatory forms.

Historical examples are used to demonstrate pharmacovigilance reasoning. Unless explicitly stated in the cited regulatory record, the article does not claim that EMA or a national competent authority formally classified a historical event as an "emerging safety issue" under the GVP definition at the time.

The purpose of the historical examples is to show how evidence, uncertainty, clinical consequence, risk-management effectiveness and regulatory urgency can interact in real pharmacovigilance decision-making.

Revision History

Related Articles

Emerging Safety Issues in Pharmacovigilance

A detailed guide to emerging safety issues, regulatory expectations, escalation procedures, QPPV oversight and inspection considerations.

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.

Benefit-Risk Evaluation in Pharmacovigilance

A comprehensive guide to benefit-risk evaluation, regulatory expectations, pharmacovigilance activities and decision-making throughout the product lifecycle.