Estimating Incidence Rate on Current Status Data Imputing Partial Status with Application to Pediatric Cancer Survivor Study

• Activity Number: 343
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Shesh N. Rai and Jianmin Pan*+ and Xiaobin Yuan
• Companies: University of Louisville and University of Louisville and University of Louisville
• Keywords: Phase IV clinical trial ; Cardiotoxicity ; Cross-section survey data ; Interval censored data ; K-M method ; Imputation

Abstract

New drug discovery has dramatically improved survival, but with long-term adverse events. This motivates the examination of adverse outcomes such as long-term drug toxicity in a Phase IV trial. An ideal approach to monitor long-term toxicity is to systematically follow the survivors, which is generally not feasible. Instead, cross-sectional surveys are sometimes conducted to investigate the long-term effects of cancer treatment. In these studies, one objective of interest is to estimate the cumulative incidence rates of toxicity with specific interest in fixed-term rates. However, such data poses many issues: incomplete data, competing risks and selection bias, etc. An example for this study is described in Hudson et al. (2007) studying the effect of cardiotoxicity due to anthracyclines exposure during cancer treatment. The main issue in such studies is the high missing rate. In this paper, we impute the missing values using multivariate regression method under some parametric assumptions, combine with methods previously described to estimate the cumulative incidence rates in an illness-death/failure model, and compare with the results obtained without using imputation.