PV Modules

Causality assessment is the clinical step in pharmacovigilance where a medically-qualified reviewer (or a multidisciplinary team for complex cases) evaluates whether the suspected medicine is likely to have caused the adverse event (AE). The outcome influences regulatory reporting, labeling, signal detection, and clinical follow-up. Below I cover purpose, principles, common methods, a step-by-step approach, examples, documentation, pitfalls, and best practices.

1) Purpose — why we do causality assessment

• Decide whether an event is possibly, probably, or unlikely related to a drug.
• Determine whether the case needs expedited regulatory reporting (e.g., SUSAR: serious, unexpected, and suspected causal relationship).
• Inform medical management (stop drug, supportive care, rechallenge decisions).
• Support signal detection and safety surveillance.
• Provide a transparent clinical rationale that can be audited.

2) Key principles and inputs used

Causality is not purely algorithmic — it’s a clinical judgement supported by evidence. Typical inputs:

• Temporal relationship: time from drug start (or dose change) to AE onset.
• Dechallenge: improvement when drug withdrawn.
• Rechallenge: recurrence on re-exposure (strong evidence if ethical/available).
• Plausibility: pharmacology, known ADR profile, class effects.
• Alternative explanations: underlying disease, other medications, infections, lab abnormalities.
• Dose–response: AE related to dose or accumulation.
• Objective data: labs, biopsy, ECG, imaging.
• Previous reports: literature, company safety database, spontaneous reporting systems.
• Quality of information: completeness and reliability of the case.

3) Common structured methods / scales

These are tools to make the process consistent. They differ in design and use case.

A. WHO-UMC System (widely used in regulatory context)

Categories: Certain, Probable/Likely, Possible, Unlikely, Conditional/Unclassified, Unassessable/Unclassifiable.
Relies on temporal relationship, dechallenge/rechallenge, alternative causes, and prior knowledge.

B. Naranjo Algorithm (questionnaire with score)

A scored checklist (e.g., +2 for previous conclusive reports, +1 for temporal relationship, etc.). Common in clinical research but less used for regulatory spontaneous reports because it can over-simplify.

C. Specialized algorithms

• RUCAM for drug-induced liver injury (DILI).
• Liverpool ADR Causality Assessment Tool for certain complex ADRs.

D. Company/Regulatory Algorithms

Many sponsors create SOP-aligned causality templates combining clinical narrative plus a chosen method (often WHO-UMC) for consistency.

4) Step-by-step practical approach (how a medical reviewer should proceed)

Step A — Review the case source documents

• Read all source documents (HCP notes, discharge summary, lab reports, autopsy, ECG, concomitant meds).
• Confirm completeness of key data: onset date/time, drug start/stop dates, dose, dechallenge/rechallenge details.

Step B — Establish temporality

• Can the timing support causation?
• Example windows: minutes–hours for anaphylaxis, days–weeks for rash, weeks–months for some organ toxicities.
• If timing is implausible → probably unlikely.

Step C — Exclude alternative causes

• Look for infections, new diagnoses, disease progression, interactions, overdose, contaminants.
• Consider co-medications that are known causes.

Step D — Examine dechallenge / rechallenge

• Positive dechallenge (improvement after stopping) supports causality.
• Rechallenge with recurrence is strong evidence — but rarely done intentionally.

Step E — Assess biological plausibility and prior knowledge

• Is the AE listed in the product label or prior literature? Is there a known mechanism?
• If unexpected but biologically plausible, it may be “possible” or “probable” depending on other evidence.

Step F — Consider objective data & lab tests

• Abnormal LFTs, eosinophilia, autoantibodies, ECG changes, biopsy results all add weight.

Step G — Decide on category & document rationale

• Use your chosen method (e.g., WHO-UMC) and write a concise justification: facts and reasoning. Avoid vague statements.

5) WHO-UMC categories — how to interpret them (practical cues)

• Certain
• Time relationship plausible.
• Cannot be explained by disease/other drugs.
• Response to withdrawal plausible.
• Definitive pharmacology or objective evidence (lab) or positive rechallenge.
• Probable / Likely
• Reasonable time relationship.
• Unlikely to be explained by other causes.
• Reasonable response to dechallenge.
• Rechallenge not required.
• Possible
• Time relationship reasonable but could also be explained by disease or other drugs.
• Information may be inadequate for stronger assignment.
• Unlikely
• Temporal relationship makes causal link improbable.
• Other explanations more likely.
• Conditional / Unclassified
• Event requires more data before assessment (pending follow-up).
• Unassessable / Unclassifiable
• Insufficient or contradictory information; cannot be judged.

6) Example case (walk through)

Case: 45-yr-old female started Drug A for hypertension on April 1. On April 5 she developed generalized itchy rash and facial swelling; presented to ER on April 6. She stopped Drug A on April 6; symptoms resolved over 3 days after antihistamines. No new infections, no other new meds. She had previously tolerated Drug B (another antihypertensive).

Assessment:

• Temporality: onset 4 days after starting — plausible for hypersensitivity.
• Dechallenge: improvement after stopping → supportive.
• Alternative causes: none identified.
• Prior knowledge: Drug A known to cause hypersensitivity rash (label).
• Rechallenge: not performed (unnecessary / unethical).

WHO-UMC assignment: Probable/Likely.
Documentation: state dates, timeline, dechallenge response, absence of alternative causes, relevant literature/label citation, final category with rationale.

7) Documentation & record-keeping (what to include)

• Final causality category and method used (e.g., WHO-UMC).
• Short written justification: key facts (dates, dechallenge/rechallenge, labs, alternative causes).
• Name and designation of assessor and date of assessment.
• Any differing opinions (if multidisciplinary review).
• If changed later (after follow-up): document the reason and date.

A typical causality note:

“WHO-UMC: Probable. AE onset 4 days after starting Drug A (2025-04-05). Symptoms improved within 3 days of stopping drug (2025-04-09). No alternative causes identified; drug labeled for hypersensitivity. Assessed by Dr. X on 2025-04-12.”

8) Regulatory implications

• If you assign suspected causality for a serious AND unexpected event → expedited reporting (SUSAR) in many jurisdictions.
• If causality is unlikely, may not require expedited submission but still retained in the safety database for aggregate review.

9) Special situations & nuances

• Multiple suspect drugs: assess each drug separately; consider interactions or additive effects.
• Class effects: if the drug is similar to others known to cause the AE, that strengthens plausibility.
• Vaccines & immune events: timelines and biologic plausibility differ (e.g., immune thrombocytopenia).
• Pregnancy exposures: consider maternal vs foetal events and timing of exposure.
• Overdose / toxicity vs idiosyncratic reactions: dose–response supports causality in toxicity; idiosyncratic reactions may occur at therapeutic dose.
• Insufficient information: mark Conditional/Unclassified and request follow-up.

10) Common pitfalls to avoid

• Over-reliance on algorithm scores without clinical review.
• Ignoring alternative explanations (e.g., underlying disease).
• Using rechallenge data when rechallenge was done inadvertently but without controls — interpret cautiously.
• Coding a causality based only on temporal association (post hoc = not necessarily causal).
• Failing to update causality after receiving follow-up data.

11) QA, training, and SOPs

• Organizations should define which causality method is standard (WHO-UMC common for regulatory reporting).
• Implement peer review or consensus for borderline/serious cases.
• Train medical reviewers with case examples and inter-rater reliability exercises.
• Periodically audit causality assessments for consistency and quality.

12) Quick checklist for the assessor

1. Are start/stop/onset dates present and clear?
2. Is the temporal relationship plausible?
3. Any dechallenge / rechallenge data?
4. Are objective tests supportive?
5. Alternative causes considered & documented?
6. Is there prior information in literature/label?
7. Which causality method used? (state clearly)
8. Final category with succinct rationale recorded.
9. Assessor details and date.

13) Short reference table