Statistical Implementation of the Diabetes Guidance for Cardiovascular Outcomes
*David Hoberman, Food and Drug Administration 

Keywords: diabetes, cardiovascular, statistics

The 2008 FDA/CDER diabetes cardiovascular outcomes guidance provides very general advice to sponsors when designing a program to assess cardiovascular risk. As a result, it has generated a wealth of protocols which present statistical challanges including meta-analyses with sparse data, heterogeneous sources of data, sequential analyses and adaptive designs, all in the context of evaluating two null hypotheses. Some protocols combine some of the above features; others nearly all. This talk will briefly summarize reviewers' experience with the expectation that questions and answers following the talk will help clarify the issues confronting the FDA's review of cardiovascular outcome submissions for drugs to treat type 2 diabetes.