Subgroup Analyses in Clinical Trials: Can We Smooth Out the Rough Edges?

Presented: December 6th, 2007

Speaker: Daniel Sargent Ph.D., Professor of Biostatistics, Mayo Clinic

View the on-demand web cast. Download instructions on how to view the web cast.

Download the web cast presentation.

Subgroup analyses in clinical trials are simultaneously essential for scientific discovery (hypothesis generation), required for ethical and administrative reasons (differential efficacy or toxicity in subsets), and vexing (extreme difficulty in interpretation).  The greater understanding of biology, in cancer in particular, is leading to greater segmentation of the population.  New agents may be tested based on biology, not tumor site, or have efficacy in only a portion of a population (defined by a biologic marker). Additionally, technology is allowing massive multiplicity in markers explored, further complicating the issue. After motivating with several examples in this talk I will present:

• Practical considerations regarding the interpretation of subgroup analyses, and
• Statistical methods based on hierarchical models (allowing an overall mean and a subgroup specific effect), and/or shrinkage estimators (allowing interactions in the model but then shrinking them) that may assist in the analysis and interpretation of subgroups questions. 

Daniel J. Sargent, Ph.D., Professor of Biostatistics at the Mayo Clinic, is the Group Statistician for the North Central Cancer Treatment Group and the director of Cancer Center Statistics at the Mayo Clinic Comprehensive Cancer Center.  Dr Sargent has extensive experience in both clinical and basic science research.  He is a member of the US GI Steering Committee, and co-chaired the Gastrointestinal Committee for the NCI Common Data Elements Project. He has extensive peer-review activities as a member and chair of a NCI Study section (Subcommittee H), membership on several NCI Special Emphasis Panels, and ad hoc reviewing for funding agencies around the world.  Dr. Sargent co-chaired a joint NCI-EORTC committee on methodology for tumor marker studies, and was a member of the FDA panel on endpoints for colon cancer clinical trials.  Dr. Sargent is a chair or member of multiple Data Safety and Monitoring Committees. Dr. Sargent has a strong record of statistical methodological development, including the area of clinical trial design. He has published papers on innovative designs for Phase I and II clinical trials and new technology based methodologies for real-time monitoring of adverse events. Recent research interests include endpoints in clinical trials and clinical trial design for tumor marker studies.  He has authored over 120 peer-reviewed manuscripts, book chapters, editorials, and letters, including first author publications in the New England Journal of Medicine, the Journal of the National Cancer Institute, the Journal of Clinical Oncology, Cancer, Controlled Clinical Trials, Biometrics, Statistics in Medicine, and Seminars in Oncology.