Testing the Equality of the Efficiencies of Two Correlated Outcome Measures in Clinical Trials

• Activity Number: 253
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Youngju Pak*+ and John Hewett and Marian Minor and Nilotpal Sanyal
• Companies: University of Missouri-Columbia and University of Missouri-Columbia and University of Missouri-Columbia and University of Missouri-Columbia
• Keywords: Clinical trials ; Efficiencies ; Clinical t Bootstrap-t confidence intervals ; Correlations ; Pain scales

Abstract

In many of clinical trials, there will be more than one outcome variables. Anderson and Chernoff defined the efficiency to identify the variable which is the more sensitive in measuring change. We have developed the two tests for testing the equality of the efficiencies of two correlated variables and applied our methods to data from a arthritis clinical trial to compare efficiencies of 3 different Womac scales. An approximate Z-test and the bootstrap-t method are suggested to test the equality of efficiencies of two different outcome variables. When both sample size and correlation between two efficiencies are large, an approximate Z-test shows fairly good power. Application to data from an arthritis clinical trial show that the Womac difficulty scale is sensitive than both Womac pain.