This presentation addresses statistical considerations that substantially influence the design, analysis, and interpretation of confirmatory clinical trials with regulatory objectives. Some of the considerations that will be discussed include missing data and its implications to robustness of results, sufficiency of information to support efficacy of a test medicine through non-inferiority to an active control, multiplicity of response variables, multiplicity of treatments, objectives for interim analyses, objectives for subgroups, issues for multi-center designs, covariance adjustment in planned primary analyses, planned integrations of studies, issues for safety data, and revisions for analysis plans. For all of these considerations, careful planning is necessary in order to have minimal bias in estimation, to avoid excessive prevalence of findings due to chance, and to have at least good power for the confirmatory demonstration of efficacy of a test medicine (through either the contradiction of a relevant null hypothesis or a convincingly located and narrow confidence interval). A critically important pre-requisite for statistical strategies to address such objectives in a regulatory environment is their explicit specification in the protocol and formal analysis plan prior to any unmasking.
Meet the speaker in Room 212 Cockins Hall at 1:00 p.m. Refreshments will be served.