Estimating Marginal Causal Effects for Observational Survival Data

In causal survival analysis, the hazard ratio is not a good marginal effect measure due to non-collapsibility. We advocate using the restricted mean survival time (RMST) difference, which is essentially a mean difference. We propose a matched design with sensitivity analysis for observational survival data, which controls observed confounding via matching and assesses unmeasured confounding through the E-value approach. We apply the proposed method to the Atherosclerosis Risk in Communities Study (ARIC) to examine the causal effect of smoking on stroke-free survival.

Speaker: Bo Lu, PhD, Professor of Biostatistics, The Ohio State University

To obtain Zoom link please email: Liz Buggs