Causality Assessment of Adverse Events in Clinical Trials: II. An Algorithm for Drug Causality Assessment

Abstract

The noise in investigator causality assessment in clinical trials can be substantially reduced by eliminating missing data, improved data quality, appropriate use of data, use of all causality factors, converting subjective judgment, and minimizing errors in assessment. An algorithm is proposed to guide the routine drug causality assessment in clinical trials. Special features of this algorithm include dealing only with treatment-emergent events, considering first the most important factors, allowing early exit, emphasizing the results of de-challenge and re-challenge, and applying the Bayesian concept. Recognizing that knowledge of drug safety profile is learned through a series of trials, known drug-event association is used in this algorithm to modify the causality rating based on results of de-challenge and re-challenge. The investigator is also asked to identify all possible etiologies at the outset, and to compare the relative likelihood of drug-causation versus an alternative etiology to form the final assessment. Drug causality assessment of adverse events is considered a part of the learning process, and improvements in assessment can lead to collection of useful data for drug safety analysis and utilization of an important source of information, the physician investigator.