An Adjusted Cumulative Incidence Function Estimator for Unbalanced Groups

• Activity Number: 420
• Type: Contributed
• Date/Time: Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Yongyun Zhao*+ and Chung-Chou H. Chang
• Companies: University of Pittsburgh
• Keywords: Adjusted cumulative incidence function ; competing risks ; cumulative incidence function ; inverse probability weighting ; ;

Abstract

Cumulative incidence function (CIF) estimates are commonly used to describe the failure probability when competing risks are present. CIF estimates can be derived from Kaplan-Meier survival and cause-specific hazard estimates. In observational studies or nonrandomized trials, the CIF estimates may be biased if the distribution of a confounding variable differs among groups. To reduce bias, we developed an adjusted CIF (ACIF) estimator that is based on the use of inverse probability weighting. To obtain the ACIF, we first determine the probability that each individual will be in a particular group, conditional on the individual's characteristics. After we assign a weight equal to the inverse of this probability, we use the weight to adjust the Kaplan-Meier survival and cause-specific hazard estimates. To illustrate, we use the example of HIV treatment groups.