Causal Interference with Time-varying Exposures
• James Robins
Mitchell L. and Robin LaFoley Dong Professor of Epidemiology
Departments of Epidemiology and Biostatistics, Harvard School of Public Health, Boston, MA, United States

In this talk I discuss methods for the estimation of the causal effect of a time-varying exposure on a disease outcome of interest in the presence of time-dependent confounders. Even when confounding by unmeasured factors and model misspecification are both absent, conventional analytic methods may be biased and result in estimates of effect that may fail to have a causal interpretation, regardless of whether or not one adjusts for the time-dependent confounders in the analysis. However an unbiased estimate of the causal effect may be obtained by using one of the three so-called g-methods: : the parametric g-computation algorithm formula (the "g-formula"), inverse probability of treatment weighting (IPTW) of marginal structural models (MSMs), and g-estimation of structural nested models. The strengths and limitations of each of the three g-methods will be discussed.

James Robins is Professor of Epidemiology and Biostatistics at the Harvard School of Public Health. The principal focus of Dr. Robins' research has been the development of analytic methods appropriate for drawing causal inferences from complex observational and randomized studies with time-varying exposures or treatments.