GVP Module VI: Causality Assessment and Clinical Evaluation of ICSRs

Introduction

Causality assessment asks whether the available evidence supports a relationship between a medicinal product and a reported clinical event.

It is important to distinguish this question from other ICSR assessments:

Is the case valid?
        ↓
Validity

Does the event meet a seriousness criterion?
        ↓
Seriousness

Is the reaction described in the applicable reference information?
        ↓
Expectedness

Does the evidence support a relationship with the product?
        ↓
Causality

A suspected adverse reaction does not require proof of causation. Pharmacovigilance operates in conditions of incomplete information, and the purpose of clinical evaluation is to assess the evidence available without converting uncertainty into false certainty.

1. Why Causality Matters

Causality contributes to the medical interpretation of individual cases and can influence broader safety assessment.

Information relevant to causality can also contribute to:

• signal detection and validation;
• aggregate safety evaluation;
• benefit-risk assessment;
• risk management;
• and regulatory decision-making.

The individual case should therefore contain an accurate clinical chronology and sufficient medical context for meaningful evaluation.

2. Causality Is Not Proof

A causal assessment is an evaluation of evidence, not a demonstration of scientific certainty.

The presence of a temporal relationship alone does not establish causality. Similarly, the absence of a positive rechallenge does not automatically exclude a relationship.

The assessment should consider the complete clinical context.

3. Temporal Relationship

The timing between exposure and onset of the event is an important component of clinical assessment.

The reviewer may consider:

• date of treatment initiation;
• dose and exposure duration;
• onset of the event;
• dose changes;
• treatment interruption;
• and subsequent clinical course.

The plausibility of the timing depends on the pharmacology, disease and clinical event involved.

4. Dechallenge

Dechallenge refers to what happens after the suspected medicinal product is discontinued or reduced.

Improvement following withdrawal can support a causal relationship in some circumstances, but it is not definitive evidence because many adverse events improve naturally or after other interventions.

The reviewer should establish whether the clinical course is compatible with the suspected relationship.

5. Rechallenge

Rechallenge occurs when the medicinal product is administered again after an earlier event.

A recurrence after rechallenge can provide important evidence in some circumstances.

However, the absence of rechallenge is common and should not automatically weaken a case to the point of excluding a possible relationship. Deliberate rechallenge may also be clinically inappropriate where a serious reaction has occurred.

6. Alternative Causes

Clinical evaluation should consider plausible alternative explanations.

These may include:

• underlying disease;
• comorbidities;
• concomitant medicines;
• infections;
• environmental exposures;
• procedures;
• or other relevant clinical factors.

The presence of an alternative cause does not automatically prove that the medicinal product was unrelated. The available evidence should be weighed as a whole.

7. Concomitant Medicines

Concomitant medicines can complicate causality assessment.

The reviewer should consider whether another medicinal product provides a plausible explanation for the event and whether interactions or combined effects are possible.

The case narrative should contain relevant information without turning every concomitant medicine into a presumed competing cause.

8. Medical History

Pre-existing disease can be highly relevant to the interpretation of an adverse event.

For example, a patient with a known cardiovascular condition may experience an event that could arise from the underlying disease as well as from medicinal-product exposure.

The medical history should therefore be considered in context.

9. Objective Evidence

Objective findings can strengthen or weaken a causal hypothesis.

Examples include:

• laboratory results;
• imaging;
• pathology;
• clinical examination;
• diagnostic testing;
• and documented treatment response.

The value of objective evidence depends on its quality, timing and clinical relevance.

10. Biological and Pharmacological Plausibility

The known pharmacology of a medicinal product can inform causality assessment.

Relevant considerations can include:

• mechanism of action;
• pharmacodynamic effects;
• pharmacokinetics;
• known class effects;
• dose-response relationships;
• and biological plausibility.

Pharmacological plausibility should support clinical reasoning rather than substitute for evaluation of the individual case.

11. Clinical Review

Clinical evaluation should integrate the available information into a coherent assessment.

The reviewer should consider the chronology, medical history, concomitant treatment, alternative causes, dechallenge/rechallenge information and objective findings together.

A good case assessment explains the evidence rather than merely assigning a label.

The next chunk will cover causality categories, WHO-UMC and algorithmic approaches, special situations, medical review, case narratives and difficult causality scenarios.

12. Causality Categories

Organisations may use defined causality categories or assessment approaches within their safety systems. The important requirement is that the approach is controlled, understood by trained personnel and applied consistently.

A causality category should not be interpreted as a statement of scientific certainty. It represents the assessment supported by the available evidence.

13. WHO-UMC Approach

The WHO-UMC causality categories are widely used in pharmacovigilance and can provide a structured framework for evaluating individual cases.

Categories such as certain, probable/likely, possible, unlikely, conditional/unclassified and unassessable/unclassifiable reflect different levels and limitations of evidence.

An organisation using this approach should ensure that personnel understand the underlying criteria rather than selecting a category based solely on intuition.

14. Algorithmic Approaches

Some organisations use structured or algorithmic causality tools.

Algorithms can improve consistency when appropriately designed and controlled, but a numerical output should not be mistaken for independent clinical judgement.

The organisation should understand the algorithm's assumptions, limitations and intended use.

15. Global and Local Assessments

A case may be assessed differently by different parties because they are answering different questions or applying different procedures.

For example, a source reporter's opinion, an affiliate assessment and an MAH medical assessment may not be identical.

The safety system should preserve relevant source information while applying the organisation's controlled assessment process.

16. Reporter Causality

The reporter's opinion can be clinically important information.

However, a reporter's attribution should not automatically become the MAH's final medical assessment. The organisation should distinguish what the reporter concluded from what the MAH independently determines from the available evidence.

17. Causality and Seriousness

Causality and seriousness are independent dimensions.

A serious event can be unrelated to the medicinal product, while a non-serious reaction can have a strong temporal and biological relationship with treatment.

The presence of a seriousness criterion should therefore not automatically increase the assessed probability of causality.

18. Causality and Expectedness

Expectedness is also independent of causality.

A reaction can be expected and causally related, expected but unrelated, unexpected and plausibly related, or unexpected with little evidence supporting a relationship.

These assessments should not be collapsed into a single classification.

19. Causality and Dechallenge

Improvement after withdrawal can support causality when the clinical course is compatible with a product-related effect.

However, spontaneous recovery, treatment of the underlying condition or other interventions may provide alternative explanations.

Dechallenge should therefore be interpreted in context.

20. Causality and Rechallenge

A positive rechallenge can be important evidence, particularly when the clinical event recurs in a compatible temporal relationship after re-exposure.

Nevertheless, rechallenge is often absent and should not be expected routinely. Deliberate rechallenge may be inappropriate for serious reactions.

21. Special Situations

Causality assessment can become particularly complex in special situations.

Examples include:

• overdose;
• medication error;
• misuse;
• abuse;
• off-label use;
• pregnancy exposure;
• occupational exposure;
• and lack of efficacy.

The reviewer should first establish what clinical event actually occurred and then assess its relationship with the medicinal product.

22. Lack of Efficacy

Causality concepts should not be applied mechanically to every lack-of-efficacy report.

The reviewer may need to consider disease progression, adherence, dose, treatment duration, resistance, underlying disease characteristics and alternative explanations.

The clinical question is different from simply asking whether a conventional adverse reaction was caused by the product.

23. Medication Error and Overdose

Where a medication error or overdose occurs, the causality assessment should distinguish the exposure circumstance from any clinical consequence.

For example:

Incorrect dose
     ↓
Exposure
     ↓
Clinical event
     ↓
Causality of clinical event

The error itself should not automatically be described as an adverse reaction.

24. Pregnancy Exposure

Pregnancy exposure may initially provide no clinical event against which conventional causality can be assessed.

If a maternal, fetal or neonatal outcome is subsequently reported, the clinical chronology and relevant exposure information should be evaluated.

Longitudinal follow-up can therefore be important to meaningful medical assessment.

25. Narrative and Causality

The case narrative should provide the evidence needed to understand the causality assessment.

A strong narrative should allow the reader to reconstruct:

• exposure;
• onset;
• clinical course;
• investigations;
• treatment;
• outcome;
• dechallenge/rechallenge;
• relevant medical history;
• and alternative causes.

The narrative should not contain unsupported conclusions presented as facts.

26. Medical Review Triggers

The organisation should define when medical review is required or advisable.

Potential triggers can include:

• serious cases;
• complex clinical events;
• significant diagnostic uncertainty;
• conflicting information;
• important new follow-up information;
• unusual patterns;
• and cases potentially relevant to signals.

The precise criteria should be controlled within the PV quality system.

27. Documentation of Clinical Reasoning

The level of documentation should be sufficient to explain material medical decisions.

Where a difficult case receives a specific causality conclusion, the organisation should be able to reconstruct the evidence considered and the basis for the assessment.

The next chunk will address difficult causality scenarios, causality reassessment, inspection findings, quality controls, practical examples, final principles, References and Regulatory Note.

28. Difficult Scenario: Strong Temporal Relationship

A reaction occurs shortly after administration of a medicinal product, but the patient has a disease that can independently produce the same event.

The temporal relationship is relevant evidence, but it should not be treated as proof of causality. The reviewer should consider the baseline disease, alternative causes, known product effects and the overall clinical chronology.

29. Difficult Scenario: Positive Dechallenge

A patient improves after the suspected medicinal product is discontinued.

This may support a causal relationship, but the reviewer should consider whether the event would have improved naturally or whether another treatment was responsible for the improvement.

30. Difficult Scenario: No Dechallenge

The medicinal product cannot be discontinued because it is essential for treatment, or the patient continues treatment despite the reaction.

The absence of dechallenge information does not automatically exclude causality. The assessment should rely on the other available evidence.

31. Difficult Scenario: Positive Rechallenge

The same reaction recurs after re-exposure to the product.

Where the chronology and clinical circumstances are credible, this can provide strong evidence supporting a relationship. However, the assessment should consider alternative explanations and whether the re-exposure was intentional, accidental or clinically necessary.

32. Difficult Scenario: Multiple Concomitant Medicines

A patient takes several medicines and develops a reaction that is compatible with more than one product.

The reviewer should assess each plausible suspect and consider chronology, known effects, dose changes, dechallenge, rechallenge and alternative causes.

The existence of several possible suspects does not justify assigning causality arbitrarily to the newest or most prominent medicine.

33. Difficult Scenario: Underlying Disease

The reported event is also a recognised manifestation of the patient's underlying disease.

The assessment should consider the natural history of the disease, disease severity, timing of treatment and any evidence that the event changed after exposure.

34. Difficult Scenario: Conflicting Source Opinions

A reporter considers the product causally related, while the MAH medical reviewer considers another explanation more plausible.

Both pieces of information should be preserved where relevant. The reporter's opinion should not be erased, but it does not automatically determine the MAH's medical assessment.

35. Reassessment After New Evidence

Causality is not necessarily a one-time decision.

New information can change the assessment, including:

• a new diagnosis;
• laboratory evidence;
• a new concomitant medicine;
• confirmed dechallenge;
• rechallenge;
• a competing clinical explanation;
• or additional epidemiological or scientific evidence.

The organisation should have a process for identifying material information that requires reassessment.

36. Case-Level Versus Signal-Level Causality

An individual case assessment and a broader signal assessment answer different questions.

A collection of cases with individually uncertain causality can nevertheless contribute to a signal when considered collectively. Conversely, a strongly suspected individual case does not by itself establish a population-level causal relationship.

The organisation should preserve this distinction in its safety processes.

37. Inspection Considerations

An inspector may ask:

• Who performs causality assessment?
• What procedure or methodology is used?
• Are personnel trained in the methodology?
• How are difficult cases escalated?
• How are reporter and MAH assessments distinguished?
• How is new information incorporated?
• How are alternative causes considered?
• How is medical review documented?

The organisation should be able to demonstrate a reproducible process rather than relying on individual undocumented judgement.

38. Inspection Risk: Checkbox Causality

A common weakness is selecting a causality category without documenting or demonstrating meaningful clinical evaluation.

A controlled assessment should be supported by the information in the case and, where appropriate, documented medical reasoning.

39. Inspection Risk: Temporal Relationship as Sole Evidence

Another weakness is treating temporal sequence as sufficient proof.

The fact that an event occurred after exposure is necessary for some causal hypotheses but is not, by itself, sufficient to establish causality.

40. Inspection Risk: Ignoring Alternative Causes

A causality assessment that never considers the underlying disease, concomitant medicines or other plausible explanations may be clinically incomplete.

The level of consideration should be proportionate to the case and the clinical question.

41. Inspection Risk: Failure to Reassess

New information may materially change the evidence but remain only in free text while the original causality assessment remains unchanged.

Quality controls should identify material follow-up requiring reassessment.

42. Quality-System Controls

A robust causality process should include:

• defined methodology;
• competent personnel;
• medical-review criteria;
• controlled terminology where applicable;
• access to relevant clinical information;
• quality control;
• escalation pathways;
• and auditability of material assessments.

The process should also explain how causality interacts with case processing, signal management and aggregate safety evaluation.

43. Practical Example: Dechallenge Supports but Does Not Prove

A patient develops a rash after starting a medicine. The medicine is discontinued and the rash resolves.

The chronology and improvement after withdrawal support a possible relationship. However, the reviewer also considers whether another medicine was stopped at the same time and whether the rash could have resolved spontaneously.

The final assessment reflects the totality of evidence.

44. Practical Example: Rechallenge

A patient develops the same clinical event after two separate exposures to the same product, with compatible timing on both occasions.

The repeated temporal relationship can provide important evidence. The reviewer documents the chronology and considers whether another factor was present during both episodes.

45. Practical Example: Alternative Cause More Plausible

A patient develops a symptom after starting a product but has a severe underlying disease that commonly produces the same symptom. There is no improvement after discontinuation and subsequent investigation supports disease progression.

The available evidence may make a product-related explanation less convincing, even though the event followed exposure.

46. What Good Looks Like

A mature causality process:

• evaluates the complete clinical chronology;
• considers alternative explanations;
• distinguishes source opinion from MAH assessment;
• uses a defined and understood methodology;
• documents material clinical reasoning;
• reassesses cases when evidence changes;
• connects individual-case assessment with broader safety surveillance;
• and remains inspection-ready.

47. Final Principles

1. Causality is an evidence-based clinical assessment, not proof of scientific certainty.
2. Temporal relationship is important but is not sufficient by itself.
3. Dechallenge and rechallenge can provide useful evidence but must be interpreted clinically.
4. Alternative causes should be considered where relevant.
5. Concomitant medicines and underlying disease can materially affect assessment.
6. Reporter causality and MAH causality should remain distinguishable.
7. Causality is separate from validity, seriousness and expectedness.
8. New evidence can require reassessment of a previous conclusion.
9. Individual-case causality and population-level signal assessment are different activities.
10. The organisation should use a controlled, reproducible and appropriately documented methodology.

Key Takeaways

• Good causality assessment is clinical reasoning supported by evidence.
• A causality category without a defensible evidence base is weak pharmacovigilance practice.
• The absence of dechallenge or rechallenge does not automatically determine the conclusion.
• The best assessment considers the whole case rather than one attractive piece of evidence.
• Follow-up can materially change causality and should trigger reassessment where appropriate.
• Individual cases contribute to, but do not by themselves determine, population-level causal conclusions.

References

1. European Medicines Agency. Good Pharmacovigilance Practices (GVP), Module VI — Collection, management and submission of reports of suspected adverse reactions to medicinal products.
2. World Health Organization. The use of the WHO-UMC system for standardised case causality assessment.
3. European Medicines Agency. GVP Module IX — Signal management.
4. International Council for Harmonisation. ICH E2D — Post-Approval Safety Data Management: Definitions and Standards for Individual Case Safety Reports.
5. European Parliament and Council. Directive 2001/83/EC, as amended.
6. European Commission. Commission Implementing Regulation (EU) No 520/2012, as amended.

Regulatory Note

This article is an educational explanation of clinical causality assessment within the EU pharmacovigilance framework. It does not replace current GVP requirements, applicable legislation, product-specific regulatory requirements, organisational procedures or competent-authority expectations.

Causality methodologies may differ between organisations and regulatory contexts. Where a specific methodology is required or adopted, the current applicable source and the organisation's controlled procedure should be followed.

The practical examples are illustrative and are not descriptions of specific regulatory inspection cases unless an authoritative source is explicitly identified.