ABSTRACT
The analysis plan is an essential part of any randomized clinical trial. Most statisticians have considerable experience with writing adequate analysis plans for studies that have one or two univariate endpoints. For example, clinical trials of treatments for cancer have a primary and one to two secondary endpoints, which are either the time-to-an-event (e.g., disease progression or death) or a binary outcome such as a complete or partial response (measured as the change in tumor size). The analysis of these univariate outcomes is straightforward.
