Evaluation of drug toxicity in clinical trials.

An increasing number of drugs removed from the market because of unacceptable toxicity raises concerns regarding pre-approval testing of drug safety. In the present paper it is postulated that the non-inferiority type of trial should be abandoned in favor of the superiority trial with active controls and less stringent (p<0.1, both for efficacy and toxicity) statistics. This approach will increase sensitivity of detection of drug-induced adverse effects at the expense of increasing false positive results regarding the difference in efficacy between the tested and reference drug. Such a move will increase the protection of future patients. In addition, the proposed design is far more acceptable from the clinical (e.g. no need to specify the statistically expected "unimportant" number of deaths) and ethical points of view, as well as being favored by the strong incentive of involved parties. In the second part of this paper arguments are presented in favor of the hypothesis that placebo (still used in some superiority trials) does not induce adverse effects. The assertion that placebo may induce adverse effects is probably biased by the nature of the clinical experiment. Such a conclusion is supported by studies indicating that placebo-induced adverse effects are disease -- and treatment -- specific. The modification of clinical trials according to the proposed changes may increase the trials' sensitivity at detecting adverse effects of drugs.