Estimation of Covariate Effects for Interval-Censored Competing Risks Data Under the Joint Modeling Framework

• Activity Number: 186
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Bo Fu*+ and Chung-Chou Chang and Ching-wen Lee
• Companies: Merck and University of Pittsburgh and University of Pittsburgh
• Keywords: Competing risks ; Informative dropout ; Interval censoring ; Joint modeling

Abstract

When the exact event time falls in between or before the observed times (e.g. survey or visit based survival data in clinical trial or observational study), then the interval censoring time-to-event method would be an appropriate option of analyzing the effects of factors on the event of interest. Informative dropout due to competing risks is one practical aspect that researchers need to take into account. Failing to account the association between the event of interest and the informative dropout might induce bias to estimate the effects of factors with respect to the main event. A joint modeling method is proposed by using common factors and random terms to capture the dependency among different causes of event for the scenario when both the event of interest and the informative dropout are interval censored data. Two likelihood structures are provided, which could be combined in practice. Simulations were performed to examine the performance of proposed method. The proposed method was applied to a study of mild cognitive impairment (MCI) to identify the relationship between vascular variables and the progression to MCI, while considering potential informative dropout to reduce t