Alternatives to Hazard Ratios for Comparing the Efficacy or Safety of Therapies in Noninferiority Studies

• DOI: 10.7326/m14-1741
• Publication Date: 2015-06-08T22:31:12Z
• Authors: Hajime Uno; Janet Wittes; Haoda Fu; Scott D. Solomon; Brian Claggett; Lu Tian; Tianxi Cai; Marc A. Pfeffer; Scott R. Evans; Lee-Jen Wei
• Keywords: Research Design

A noninferiority study is often used to investigate whether a treatment's efficacy or safety profile acceptable compared with an alternative therapy regarding the time clinical event. The empirical quantification of treatment difference for such routinely based on hazard ratio (HR) estimate. HR, which not relative risk, may be difficult interpret clinically, especially when underlying proportional hazards assumption violated. precision HR estimate depends primarily number observed events but directly exposure times sample size population. If event rate low, require impractically large ensure that prespecified criterion attainable. This article discusses deficiencies in current approach design and analysis study. Alternative procedures are provided, do depend any model assumption, compare 2 treatments. For study, patients' more clinically important than events. long enough evaluate reliably, then these can effectively provide interpretable evidence safety, even relatively few These illustrated data from studies. One explores cardiovascular pain medicine; second examines new diabetes. strategies intervention lead meaningful interpretations results conventional strategy uses