Primary Submission Category: Machine Learning and Causal Inference
Doubly Robust Estimation of Treatment Effect for Time-to-event Outcome under Dependent Left Truncation and Informative Right Censoring
Authors: Yuyao Wang, Andrew Ying, Ronghui Xu,
Presenting Author: Yuyao Wang*
In aging studies or prevalent cohort studies, causal inference for time-to-event outcomes can be challenging. The challenges arise because, in addition to the potential confounding bias from observational data, the collected data usually also suffers from the bias by informative right censoring and the selection bias by left truncation, where only subjects with time to event (such as death) greater than the enrollment times are included. To assess the treatment effect on time-to-event outcomes in such settings, inverse probability weighting (IPW) is often employed. However, IPW is sensitive to model misspecifications, which makes it vulnerable, especially when faced with three sources of biases. Moreover, IPW is inefficient. To overcome these issues, we construct a model doubly robust estimator that have protection against the model misspecifications for all three sources of missing mechanisms, as well as a rate doubly robust estimator that is root-n consistent even when slower than root-n methods, such as nonparametric or machine learning methods, are incorporated. Our work represents the first attempt to construct doubly robust estimators that account for all three sources of biases: confounding bias, selection bias from covariate-induced dependent left truncation, and bias from informative censoring. We apply the proposed estimator to analyze the effect of midlife alcohol consumption on late life cognitive impairment using data from the Honolulu Asia Aging Study.