IEU Seminar - Stijn Vansteelandt from Ghent University
IEU Seminar - Stijn Vansteelandt from Ghent University
Title: Instrumental variable estimation of time-to-event endpoints
Abstract: Bias due to unobserved confounding can seldom be ruled out with certainty when estimating the causal effect of a nonrandomized treatment. The instrumental variable (IV) design offers, under certain assumptions, the opportunity to tame confounding bias, without directly observing all confounders. The IV approach is very well developed in the context of linear regression and also for certain generalized linear models with a nonlinear link function. However, IV methods are not as well developed for regression analysis with a censored survival outcome. In this talk, I will review recent developments on IV estimation of time-to-event endpoints under additive hazards models as well as Cox models. I will discuss two-stage methods (Tchetgen Tchetgen et al., 2015) that have the advantage of being simple, but the disadvantage of demanding parametric assumptions on the distribution of the exposure, amongst others. I will additionally discuss more efficient g-estimation approaches under additive hazards models (Martinussen et al., 2017) and Cox models (Martinussen, Sorensen and Vansteelandt, 2018) that are less demanding in terms of assumptions. Formal conditions are given justifying each strategy, and the methods are illustrated in a novel application to a Mendelian randomization study to evaluate the effect of diabetes on mortality using data from the Health and Retirement Study. In that context, I will also discuss how to accommodate the problem of left truncation and survivor bias which affects many Mendelian randomization studies of time-to-event endpoints (Vansteelandt, Walter and Tchetgen Tchetgen, 2018; Vansteelandt, Dukes and Martinussen, 2018).
All welcome