ABSTRACT
For the approval of a drug product, the United States Food and Drug Administration (FDA) requires that at least two adequate and well-controlled clinical trials be conducted in order to provide substantial evidence of the effectiveness and safety of the drug product. The characteristics of an adequate and well-controlled clinical trial include a valid design and appropriate statistical tests for data analysis. A valid statistical design can not only minimize bias and variability that may be associated with the trial but also help to address the scientiŒc/medical questions and/or hypotheses of the trial. An appropriate statistical test can provide a fair and unbiased assessment of the effectiveness and safety of the study drug with certain assurance. When conducting a clinical trial, it may be desirable to have the study done at a single study site if (1) the study site can provide an adequate number of relatively homogeneous patients that represent the targeted patient population under study and (2) the study site has sufŒcient capacity, resources, and supporting staff to sponsor the study. The advantage of a single-site study is that it provides consistent assessment for the efŒcacy and safety in a similar medical environment. However, a single-site study has some limitations and hence may not be feasible in many clinical trials, especially when the intended clinical trials are for relatively rare chronic diseases and the clinical endpoints for the intended clinical trials are relatively rare (Goldberg and Kury, 1990). As an alternative, a multicenter trial is usually considered. A multicenter study is a study conducted at more than one distinct center where the data collected from these centers are intended to be analyzed as a whole. Unlike a single-site study, a multicenter trial is much more complicated. Although, in practice, multicenter trials do expedite the patient recruitment process, some practical issues in design and analysis need to be carefully considered. These design and analysis issues include the selection of centers, the randomization of treatments, the use of a central laboratory for laboratory evaluation, and the existence of treatment-by-center interaction (Chow and Liu, 1998b). Note that the FDA indicates that an a priori division of a single multicenter trial into two studies is acceptable to the FDA for establishing the reproducibility of drug efŒcacy to new drug application approval.
