A Semiparametric Bayesian Approach for Instrumental Variable Analysis with Arbitrarily Censored Time-to-Event Outcome

• Activity Number: 608
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Xuyang Lu* and Gang Li and Hulin Wu
• Companies: UCLA and UCLA Fielding School of Public Health and University of Rochester
• Keywords: instrumental variable analysis ; time-to-event outcome ; semiparametric Bayesian model ; Dirichlet process mixtures ; Mendelian Randomization ; accelerated failure time model

Abstract

Instrumental variable (IV) analysis has been widely used in economics, epidemiology, and other fields to estimate the causal effects of covariates on outcomes, in the presence of unobserved confounders and/or measurement errors in covariates. However, IV methods for time-to-event outcome with censored data remain underdeveloped. We propose a semiparametric Bayesian approach for IV analysis with time-to-event outcome, by using a two-stage linear model with Dirichlet process mixtures for the random errors, in order to relax the parametric assumptions and address heterogeneous clustering problems. Our method applies to arbitrary censoring patterns, including left censoring, right censoring, interval censoring, or any combination of these. A Markov Chain Monte Carlo sampling method is developed to estimate the endogenous parameter. The performance of our method is examined by simulation studies. Compared with the method that ignores the unobserved confounders and measurement errors, our method largely reduces bias and greatly improves coverage probability of the estimated causal effect. In addition, compared with a parametric Bayesian IV model with bivariate normal error distribution