Do Subgroup Analyses Harm? Usually Not. Are They Helpful? It Depends. A Regulatory and Practical Review

• Activity Number: 285
• Type: Invited
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Christoph Muysers*
• Companies: Bayer HealthCare
• Keywords: clinical trials ; subgroup analysis ; guideline review ; practical approaches ; consistency

Abstract

Prospective planned confirmatory subgroup analyses play already an important role in clinical trials when the aim is a specific population claim. For such claims, numerous statistical methods are available to consider multiplicity issues, adaptive and enrichment design features in order to provide finally reliable inference for a certain subgroup. Another aim of subgroup analyses is to ensure consistency of the response within the final target population. Well known representatives for that purpose comprise among others interaction terms, forest plots, or even pure stratified repetitions of the primary analysis. Within the regulated environment, there is an increasing demand to formalize the use of these consistency checks. It is a joint goal of statisticians from health authorities and industry to reduce post-hoc discussions on post-hoc analyses and thus to agree upon a reasonable approach already in the planning phase. Whereas a forced consideration of consistency is appreciated, it is difficult to come up with one reasonable approach fulfilling all global requirements in one analysis plan. This presentation reflects current practice and available guidance and highlights issues.