Statistical aspects of clinical trials of antibiotics in acute infections.

Controlled clinical trials are important tools for evaluating antibiotics in acute infections. External and internal validity, definition of efficacy criteria, and size of the patient sample constitute special statistical problems in such studies. Critical issues regarding external validity pertain to the selection of patients and to the concept of consecutive patients. The internal validity of a study is influenced by the withdrawal of patients from the evaluation after randomization and the comparability of treatment groups with regard to prognostic factors. The definition of efficacy criteria on the basis of bacteriologic outcomes across control visits is not straightforward. Particularly, the evaluation of efficacy at the last follow-up visit must take into account the accumulated information rather than the cross-sectional information. The most common situation in comparative trials of antibiotics is that rather small differences in efficacy can be anticipated. Sometimes, the question at issue is the demonstration of antibiotic equivalence. For valid conclusions to be made in such situations, large samples must be used. A basic problem affecting many studies of antibiotics is that this criterion is not fulfilled.