Causality Assessment

Causality assessment converts a reported adverse event into a graded statement about attribution. The inputs are consistent across methods: temporal plausibility between exposure and onset, whether the event resolved on stopping, whether it returned on restarting, whether an alternative explanation such as intercurrent illness or a concomitant medicine fits at least as well, whether the event is consistent with known pharmacology, whether it tracked dose, and whether any objective measurement supports it.

Two instruments dominate practice. The WHO-UMC system sorts cases into certain, probable, possible, unlikely, conditional or unassessable according to how many criteria are satisfied. The Naranjo algorithm scores ten questions and converts the total into definite, probable, possible or doubtful, with a positive rechallenge carrying heavy weight. Both are deliberately crude, and their real function is to make an assessor state which criteria were met rather than assert a conclusion. Reporting duties follow: an event that is serious, unexpected and at least possibly related triggers expedited submission.

Causality is judged per case, not per drug. It is what converts an adverse event into an adverse drug reaction in the pharmacovigilance sense, and where reporter and sponsor disagree, both assessments travel with the report. Dechallenge and rechallenge carry the most weight, yet deliberate rechallenge is rarely ethical, so the strongest available evidence is usually unobtainable by design.

The failure to guard against is treating a well-graded single case as a population risk estimate. A probable attribution in one patient says something about that patient and almost nothing about incidence. Attribution collapses entirely when several unapproved compounds are used at once, which is the norm in self-experimentation.