Instrumental Variable Additive Hazard Models

• Activity Number: 10
• Type: Invited
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Jason Fine*
• Companies: The University of North Carolina
• Keywords: causal ; time-to-event

Abstract

Instrumental variable methods are increasingly used in non-experimental studies to estimate the causal effects of medical interventions in the presence of unmeasured confounding. Despite the increasing use of these methods, there have been few extensions of IV methods to censored data problems. We discuss instrumental variable estimation in an additive hazards model for right censored data. Assuming linear structural equations models, a simple two stage least squares estimator is developed, with rigorous theoretical justification and inference. The methods are illustrated with data from a comparative effectiveness study of chemotherapy for colon cancer.