Chi-Square Test for Competing Risks Data

• Activity Number: 253
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Qing Yang*+ and Gang Li
• Companies: University of California at Los Angeles and University of California at Los Angeles
• Keywords: Competing risks data ; Survival analysis ; Clinical trial design ; Cox proportional model

Abstract

In many clinical trials, we may encounter correlated risk factors, which we call competing risks. In this case, even the treatment targets on one specific risk factor, it may affect other risk factors, leading to either longer or shorter overall survival time. Since both cause specific hazard and overall survival time are of great importance to evaluate the treatment effects, we want to treat them as co-primary outcomes. In this paper, we proposed a chi-square test statistic which is constructed based on two score statistics from cox proportional model. In the competing risks situation, it may be more proper than the weighted sum of linear rank statistic since it is very common to have opposite directions of the hypothesis for the competing risks data. Compare to the Bonferroni adjustment, the new test statistic is more powerful when the two outcomes are highly correlated.