Software for Group-Sequential Trials

DESCRIPTION (provided by applicant): The term "group-sequential" is used to describe clinical trials in which statistical analyses will be conducted at various points over the course of the trial. If it can be shown at an interim analysis that the treatment is effective, there may be an ethical imperative to terminate the trial. Both the NIH and FDA mandate this type of trial for certain studies, including many in the field of heart disease, cancer, and AIDS. The goal of this project is to develop a comprehensive statistical software package for the design and analysis of group-sequential trials. It will provide modules for the design stage and the interim analysis stage. During the design stage, the user will be able to choose from a wide variety of stopping rules, and to calculate sample size and power. During the interim analysis stage, the user will be able to plot the current level of evidence in relation to the pre-specified boundary, as well as calculate conditional power. It will differ from what is currently available in several respects. First, it will use realistic survival models to accurately determine sample size and power. Other software force the user to assume that hazard rates are constant for the duration of the trial. In contrast, this program will allow the hazard rates and treatment effect to vary over time. In addition, it will allow researchers to account for such factors as non-compliance, and loss to follow-up and competing risks, etc. This will substantially improve estimates of sample size and power. It will also yield better estimates of the amount of information likely to be available at each of the interim analyses, which is crucial to be able to plan the timing of the DMC meetings and total study duration. It will provide tables and graphs for optimizing designs.