The design of clinical trials with antibiotics.

Clinical trials to test antibiotics are normally of a less than satisfactory quality. Examples of common errors in the design and execution of trials with antibiotics are inclusion of too few patients to allow conclusions to be drawn, open uncontrolled design leading to a risk of considerable investigator bias, and inclusion of heterogeneous patient samples. This article suggests minimum requirements for antibiotic studies. Guidelines for such trials are needed and should have the widest possible international application in order to permit studies involving centres in various countries and to reduce the costs for development of new drugs.