Rating the severity of the medical consequences of drug-induced liver injury

Abstract

Risk analysis in drug development aims to allow for clear decisions showing whether or not the benefit of an intervention outweighs the risk. One of the difficulties in doing this in a repeatable and clear way is the problem of comparing different adverse events, as seriousness is often subjective. Using drug-induced liver injury as our model, we show that clinical, laboratory, and histological manifestations of liver reactions can be ranked by experienced hepatologists and these rankings can be used to rank the consequences of drug-induced side effects as a continuum. This risk ranked information could be transformed to standardized scores ( z score) and the risks displayed by standard techniques; adverse events can then be compared with effects from other drugs and possibly with the consequences due to untreated disease or natural occurrences. As a risk is a function of both the seriousness of the event and the probability of its occurrence, risk can therefore be displayed in terms of probability and hazard to further ease communication. We propose that risk management of drugs in development would be improved, especially in terms of risk communication, if the hazard were ranked by means of a common scale and displayed in graphic form against the likelihood of occurrence.