Resumo

LATOUR-PEREZ, J.  e  CABELLO-LOPEZ, J.B.. Early interruption of clinical trials: too good to be true?. Med. Intensiva [online]. 2007, vol.31, n.9, pp.518-520. ISSN 0210-5691.

The interruption of a randomized clinical trial when the experimental treatment is clearly superior is an ethical obligation. However, performing multiple intermin statistical analyses and the early interruption of the trial increase the possibility of alpha error, favor the overestimation of the effect, and facilitate the introduction of ineffective therapies data monitoring. Currently available empirical evidence suggests that early interrupted clinical trials overestimate the effects of new treatments and that the reasons used to justify interrupting the trial are often not sufficiently specified. This underlines the need for truly independent data monitoring committees and recommends the use of predefined restrictive rules for the interruption of a trial. Furthermore, clinicians should provide themselves with an adequate dose of skepticism and critical reading skills to enable them to explore some key aspects of these studies.

Palavras-chave : randomized clinical trials; data monitoring committees; intermediate analyses; statistical error; ethics.