Activity Number:
|
431
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 2, 2016 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #319491
|
|
Title:
|
Adjusting for Time-Varying Confounding in Survival Analysis Using Structural Nested Cumulative Survivals Models
|
Author(s):
|
Oliver Dukes* and Stijn Vansteelandt and Shaun Seaman and Torben Martinussen
|
Companies:
|
Ghent University and Ghent University and University of Cambridge and University of Copenhagen
|
Keywords:
|
Aalen's additive model ;
Causal effect ;
G-estimation ;
Marginal structural model ;
Survival data ;
Time-varying confounding
|
Abstract:
|
Time-varying confounding forms a pervasive problem in observational studies that attempt to assess the total effect of a time-varying exposure on a survival outcome. Standard regression adjustment for such confounders eliminates indirect effects of early exposures and may induce a 'collider-stratification' bias. Inverse probability weighting for marginal structural survival models overcomes these problems, but can yield unstable inferences and cannot incorporate effect modification by time-varying covariates. We show how these complications can be remedied via novel G-estimation strategies when effects are parameterized on the additive hazard scale. The proposal explicates a close link with inference under structural nested cumulative failure time models, but yields a different class of models, and estimators which accommodate continuous time settings and can be calculated in closed-form. The proposed approach naturally accommodates independent censoring and thus avoids the problem of artificial censoring to which existing G-estimation approaches are susceptible.
|
Authors who are presenting talks have a * after their name.