A0486
Title: Semiparametric efficient G-estimation with invalid instrumental variables
Authors: BaoLuo Sun - National University of Singapore (Singapore) [presenting]
Zhonghua Liu - The University of Hong Kong (Hong Kong)
Eric Tchetgen Tchetgen - The Wharton School, University of Pennsylvania (United States)
Abstract: Mendelian randomization leverages one or multiple genetic markers as instrumental variables for causal inference in the presence of possible unmeasured confounding. In order to improve efficiency, multiple genetic markers are routinely used, leading to concerns about bias due to possible violation of exclusion restriction of no direct effect of any instrument on the outcome other than through the exposure in view. To address this concern, we introduce a new class of g-estimators that are guaranteed to remain consistent for the causal effect of interest provided that a set of at least $k$ out of $K$ candidate instrumental variables are valid, for some $k$ less than or equal to $K$ set by the analyst ex-ante, without necessarily knowing the identity of the valid and invalid instruments. We provide formal semiparametric theory supporting the results and characterize the semiparametric efficiency bound for the exposure causal effect which cannot be improved upon by any regular estimator with our favorable robustness property. Both simulation studies and applications to the UK Biobank data demonstrate the superior empirical performance of our estimators compared to competing methods.
