Instrumental Variable Methods for Censored Data

• Activity Number: 224 - Recently Developed Survival Analysis Methods with Applications to End-Stage Renal Disease Data
• Type: Topic Contributed
• Date/Time: Monday, July 31, 2017 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Douglas Lehmann* and Yun Li and Douglas Earl Schaubel
• Companies: University of Miami and University of Michigan and University of Michigan
• Keywords: instrumental variable ; causal inference ; weighting ; survival analysis ; nonparametric ; observational data

Abstract

Estimation of the causal effect of treatment using observational data is often hampered by unmeasured confounding. Instrumental variable (IV) methods are gaining popularity in epidemiologic studies as a means of avoiding such bias. However, few IV methods are available for the analysis of censored data. We propose IV methods for estimating treatment effects in the presence of unmeasured confounders. After appropriately reweighting the observed risk sets, the causal effect is obtained by computing nonparametric survival functions (and the area between). A distinguishing feature of the methods is the accommodation for IV-outcome confounding. Theoretical properties of the proposed estimators are derived, with simulations used to assess finite-sample performance. The proposed methods are then applied to data from a national end-stage renal disease registry.