Planning a Randomized Trial

摘要

Randomized clinical trials are generally considered to produce results with the highest level of evidence. The study design can eliminate confounding bias by allocating patients randomly and thereby avoid systematic unbalance in confounding factors between treatment(s) and control groups. While adjustment for confounding can be made for known or suspected confounding factors in observational studies, randomization also protects against unknown confounding factors. This makes it possible to achieve valid results with optimal statistical precision. In observational studies, aspects of validity and statistical precision are often conflicting, and validity is achieved at the expense of precision. The methodology of randomized trials is complex and covers several scientific areas. This note and the one that follows present only a small number of statistics-related issues to be considered when planning and analyzing randomized trials or when reading their reports. National and international legislation may apply to clinical trials. This is not covered here.