Causality assessment in individual cases is a radically different matter, as it can easily turn into an endless argument of pros and cons of a relationship between a drug and an adverse reaction.
For single case reports, every case is assessed separately, and the evaluation can range from ‘definitely related’ through ‘probable’, ‘possible’, and ‘unlikely’ to ‘not drug-related’.
Definitely related: (> 95% confidence in causality)
- Event with plausible time relationship
- No other explanation –disease or drugs
- Positive dechallenge
- Response to withdrawal plausible
- Event definitive “pharmacologically or phenomenologically”
- Key feature: Positive rechallenge
Examples: Definitely Related
- Dizziness 3⁄4 hour after ingestion of an oral antihypertensive drug with no concomitant drugs. AE stops on stopping drug (positive dechallenge) & restarts when taken again (positive re- challenge)
- Injection site reaction 30 seconds after a subcutaneous injection
- A large tablet gets stuck in the pharynx (obstruction) while swallowing it & it has to be removed in the emergency room.
2. PROBABLE (50% to 95% confidence in causality)
- Event with plausible time relationship to drug intake
- No other explanation
- Response to withdrawal (dechallenge) clinically reasonable
- No rechallenge, or result unknown
- Key feature: Positive dechallenge
Examples: Probably Related
- Thrombocytopenia after taking an oncology drug
- Diarrhea after ampicillin
- Vaginal candidiasis after an antibiotic for bronchitis
3. POSSIBLE – (5% to 50% confidence in causality)
- Event with plausible time relationship to drug intake
- Could also be explained by disease or other medicines
- Information on drug withdrawal lacking or unclear
- Key feature: other explanations for the event are possible
Example: Abnormal liver function tests after taking an antihistamine
4. UNLIKELY ( <5% but not 0% confidence in causality)
- Event with a duration to onset that makes a relationship improbable
- Diseases or other drugs provide plausible explanations
- Event does not improve after dechallenge
- Key feature: several factors indicate strongly that the event is not a reaction
Examples: Unlikely Related
- Cancer of the colon diagnosed after 3 doses of an antibiotic
- Myocardial infarction 3 weeks after taking a drug that has a terminal half life of 10 minutes
- Auto accident but…
- – Did the drug cause dizziness which caused the accident?
5. UNCLASSIFIED/UNASSESSABLE (conditional) – (insufficient case data)
- An adverse event has occurred, but there is insufficient data for adequate assessment and additional data is awaited or under examination
- Nature of event makes it impossible to attribute causality (needs epidemiological studies)
- Key feature: Can’t assess with the information available, Data elements concerning the event are inadequate and will not be available
6. Causality ruled out/Unrelated: Events occurred prior to drug administration or events which cannot be even remotely related to drug administration. Be cautious about dismissing uncommon, SAEs that don’t seem plausibly drug-related.
Examples of SAEs that do not seem plausibly related to the drug but were found to be:
- Tendon rupture associated with the quinolone antibiotics
- Heart valve lesions associated with fenfluramine
- Retroperitoneal fibrosis with Sansert
- Pulmonary hypertension with Aminorex (a European weight loss drug), and various other drugs
- Suicidal ideation with interferons, Accutane
- Intussusception with rotovirus vaccine
- Pulmonary fibrosis with amiodarone
Ways to Determine Causality:
In general there are three approaches: unstructured or conventional, semistructured, and standardized.
1. Unstructured approach: It is based on the medical experience and knowledge of the assessor, who exercises judgement in a completely unstructured way after considering the information contained in the case report. If the judgement is not supported by a detailed discussion of the case, the grounds on which it was reached will not be clear, yet it is the most authoritative form of assessment. It is paradoxical that the most authoritative form of assessment is left to the completely subjective opinion of an assessor.
2. Semistructured approach: It provides for every causality level a descriptive and more or less loose list of what should and what should not be in the case report to assign it to a given causality level. The semistructured aproach shows how assessment was reached, even if the rules are not very specific and are mostly qualitative. World Health Organization proposed rules are one of the semistructured approach.
WHO Methodology
- Certain – Good timing, no other cause, withdrawal response plausible, rechallenge, “definitive”
- Probable – Good timing, other cause unlikely, withdrawal
- Possible – Good timing, other causes possible
- Unlikely – Poor timing, other causes more likely
- Unassessable – Insufficient or contradictory information
3. Standardized assessment: It consists of a set of questions and decision rules which result in the same answers always leading to the same final assessment.
Numerous standardized methods have been described. They are distingushed mainly by the specificity and number of items of information taken into consideration, and the weight attached to different items.
Ex: Algorithms: Use of a formal, defined mechanism or decision tree to come to a conclusion
–Imputability (France), Roussel-Uclaf (France), Venulet (Switzerland), Karsh-Lasagna (US), WHO (Sweden), Naranjo (Canada)
– Probablistic, Baysian analysis & other “statistical methods”
• Generally require more data than is available or data that is “introspective” – not yet practical.
NARANJO Algorithm: For assessing the causality-
- definite = >9
- probable = 5-8
- possible = 1-4
- doubtful = <0
Finally causality assessment is not always easy, there are difficulties in assessing AE Causality:
- Incomplete information
– Reporter may see only one of a particular AE whereas the sponsor or HA may see many from other sources & drugs - Multiple drugs taken (polypharmacy)
- Variability of clinical responses
- Underlying illness mimics AE
- Intercurrent illness
- Different medical training or viewpoint
One case example of tricky causality assessment:
A pregnant woman takes Diethylstilbestrol (DES) early in her pregnancy to prevent a miscarriage (spontaneous abortion). Fourteen years later her daughter develops cancer of the vagina.
This is the only information available in case. With this information we can assume that event is unrelated to drug. Later a study identified DES as a cause of a rare vaginal cancer in girls and young women who had been exposed to DES before birth (in utero).
