Title: An Instrumental Variable Procedure for Accounting for Unmeasured Confounding in Cox Proportional Hazard Regression Models
Abstract:
Two-stage instrumental variable procedures (estimators) are commonly used for estimating causal effects in the presence of an unmeasured confounder. In the context of the Cox Proportional Hazard regression model, this problem has recently received attention with several methods being proposed. In this talk, I develop a new estimator that is motivated by recognizing that the incumbent two-stage residual inclusion procedure induces an individual frailty in the second stage estimation procedure. The estimator is shown to perform significantly better than incumbent procedures even when the assumed distribution of the frailty is incorrect. The new estimator is used to estimate the effect of carotid endarterectomy versus carotid artery stenting on the time to death of patients suffering from carotid artery disease using a large data set from the Vascular Quality Initiative (VQI) Registry linked to United States Medicare data. If time permits, the extension of the estimator under proportional hazards to a special case of non- proportional hazards will also be presented and illustrated on the VQI data.
Biography
James O’Malley, MS, Ph.D., is Professor of Biostatistics in The Department of Biomedical Data Science and The Dartmouth Institute of Health Policy and Clinical Practice at the Geisel School of Medicine at Dartmouth. In 1999 he received his Ph.D. in Statistics from the University of Canterbury, New Zealand and a MS degree along with the L. J. Cote award for excellence in Applied Statistics from Purdue University, USA. His methodological interests encompass causal inference using instrumental variables, social network analysis, multivariate hierarchical models, and Bayesian inference with much of his work is motivated by problems in health services research. He has published over 170 peer-reviewed research papers, was chair of the Health Policy Statistics Section (HPSS) of the American Statistical Association (ASA) in 2008 and co-chaired its International Conference in 2011. In 2011 he received the HPSS Mid-career Excellence award, in 2012 was elected to be a fellow of the ASA, and was the 2019 ISPOR (International Society for Pharmacoeconomics and Outcomes Research) Research Excellence Award Recipient in Methodology.
