Guide to Causality Assessment

Mar 20, 2019
Bart Cobert

医薬品安全監視、医薬品安全性および規制関連業務に関する執筆者・専門家

In my past columns over the years I have often lamented that academia (medical, nursing, pharmacy schools) have not really gotten into the field of drug safety other than via epidemiology. Very little academic research has been or is being done in this field. In fact, as we move towards more flexible study design with the aim of studying fewer patients to demonstrate drug efficacy before approval, we will be losing safety data. Unlike efficacy, which is often statistically base, drug safety requires fairly large numbers of patients to understand the safety profile and find infrequent ADRs. We would very much like to see drug safety and pharmacovigilance become an active area of research, study and training for health care practitioners.

This posting has been prepared with Pierre Biron MD, retired Professor of Pharmacology at the University of Montreal, Canada. Prof. Biron is well known in the PV field and has authored many papers on the subject as well as several books on the topic (some of which I co-authored). This posting summarizes the information and documentation that should be examined in the study and work up of ADRs to determine whether the reaction is likely due to the drug or not. One might say it is a semi-quantitative way of categorizing an ADR in terms of causality. Such a method of describing the strength of the drug causality of the adverse event would be very useful in product labeling. This would allow the prescriber and patient to judge the likelihood of the ADR being due to the drug and help decide what work-up should or should not be done to find other possible causes.

Have a look. We’d be very interested in readers’ opinions!

Previous Documentation

1.Suspected/Suspect product

  • Is it recognized as being capable of causing the reaction, i.e. 
is the ADR mentioned in the medical literature, the official labeling or the manufacturer’s files (see proposed scale below)?
  • Is the ADR known to be frequent with this product?
  • Is the ADR pharmacologically predictable? (exaggeration of its 
pharmacologic effect, side / side effect)
  • Are there predisposing factors related to product exposure? 
(overdose, mode of administration, medication error)

2.Adverse Event


  • Is it often drug-related in general? (e.g., aplastic anemia, bullous toxidermia, torsades de pointes, fixed-drug eruption, neuroleptic malignant syndrome, acute liver failure).
  • Is it often specifically related to the suspect drug? (e.g., statins and rhabdomyolysis, SSRIs and the serotoninergic syndrome, anti-HIVs and lipodystrophy, isotretinoin and congenital anomaly, amiodarone and corneal cat’s paw deposits)
  • Are the nonchronologic clinical or laboratory features of the AE generally
suggestive of a drug reaction? (e.g., drug-induced hepatitis (e.g. Hy’s Law), lupus-like syndrome)
  • Is the reaction located in situ? Site of application (skin…), transit (esophagus), concentration (kidney…), excretion (lithiasis…).

Patient’s History

3.Prior experience specific to the ADR 

  • Prior exposure (prechallenge) to the suspect drug? If yes, was there a reaction and which one was it? (prechallenge n/a, positive, or negative)
  • Prior occurrence of the AE? If yes, in which circumstances did it occur?

4.Pharmaceutical history


  • Concomitant medication? If yes, are there any potential suspects apart from the suspect drug ? Is there a potential for a drug–drug interaction?

5.Medical history: the concomitant morbidity

  • Does the indication for the suspect drug, or other concomitant pathologies, represent potential etiologic alternatives for the AE (confounding factors)?
  • Are predisposing factors present related to the patient’s condition? (e.g., renal failure, allergy, age, sex, weight…)

Clinical Presentation

6.Chronology

  • Time to onset : since first dose, first dose then increased dose, last 
dose, single dose. A short time to onset is often very suggestive but its duration should be plausible and consistent with the pharmacodynamics and the kinetics of the suspect product. 
This is crucial information and the facts should be double-checked (patient, prescriber, pharmacist, legal guardian, nurse or other person \ as witnesses)
  • Duration of AE before stopping or reducing the dosage or after a single dose
  • Course upon dechallenge (n/a, positive, or negative). A clearly positive dechallenge is a sound argument favoring causality
  • Course upon rechallenge (n/a, positive, or negative). A positive rechallenge is one of the features with the most evidenciary value
  • Response to corrective treatment.

Proposed scale for ranking the prior documentation concerning the AE and the suspect drug:

Level 4 – Fully labeled
Level 3 – Recognized but not yet fully labeled
Level 2 – Anecdotal or predictable
Level 1 – Unpublished and unpredictable
Level 0 – Unreported worldwide (from manufacturer’s and WHO/UMC Vigibase international ADR database)

Proposed scale for quantifying causality assessment:

Level 4 – Definite ( > 95% confidence in causality)
Level 3 – Probable (50% to 95% confidence in causality)
Level 2 – Possible (5% to 50% confidence in causality)
Level 1 – Unlikely, doubtful ( <5% but not 0% confidence in causality)
Level 0 – Causality assessment impossible (insufficient case data)
Level -1 – Causality ruled out (after reviewing the case data)

For regulatory purposes in most jurisdictions, levels 1 to 4 are usually ranked as “possibly related” or having “a reasonable possibility.” In civil lawsuits, levels 3 and 4 may be deemed sufficient to infer causality (“more likely than not”). In criminal cases, even more than level 4 may be required, such as 99.9% confidence in causality.

Tags: , , , ,