Global Summary of Treatment Impact Using Multiple Patient Outcomes

• Activity Number: 471
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Brian Claggett*+ and Lu Tian and Sihai Dave Zhao and Lee-Jen Wei
• Companies: Harvard University and Stanford University and Harvard School of Public Health and Harvard University
• Keywords: clinical trials ; multiple outcomes ; ordinal logistic regression ; survival analysis

Abstract

The impact of treatment in a randomized clinical trial may manifest itself with respect to multiple patient outcomes. Typically, a single primary outcome is required, forcing investigators to select from among the possible outcomes or create a composite outcome combining some or all outcomes of interest. The selection of a single outcome ignores useful information from other outcomes. A composite endpoint may be ``dominated' by the most frequent event, which may not be of primary clinical importance. When the presence or absence of such outcomes can be used to define a set of ordered patient health states, we propose methods which take advantage of all available patient information while respecting the relative severity of such outcomes. We integrate ideas from non-parametric and semi-parametric survival methods as well as ordinal regression and propose methods for estimation, hypothesis testing, and inference for this measure of overall clinical impact of therapy. We illustrate the proposed procedure with an example from a recent cardiology trial. Though the original primary outcome failed to achieve statistical significance, our method detects a positive treatment effect.