|
Activity Number:
|
645
- Causal Mediation Analysis in Advanced Settings: Longitudinal, High-Dimensional, Censored Mediations
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 3, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Mental Health Statistics Section
|
|
Abstract #322977
|
|
|
Title:
|
Causal Mediation Analysis for a Proportional Hazards Model with Interval-Censored Failure Time Data
|
|
Author(s):
|
Wei Wang*
|
|
Companies:
|
Division of Biostatistics, CDRH, FDA
|
|
Keywords:
|
causal mediation analysis ;
natural indirect effect ;
interval-censored ;
proportional hazards model ;
potential outcome framework ;
piecewise linear approximation
|
|
Abstract:
|
The goal of causal mediation analysis is to examine the pathways and explicate the mechanisms underlying disease onset and progression in epidemiological research. The prevalence of interval-censored data is increasing in prospective cohort studies due to the periodic monitoring of the progression status. We defined the total effect, natural indirect effect, and natural direct effect on different scales (survival probability, hazard function, and restricted mean survival time) for a proportional hazards model with interval-censored data using the potential outcome framework within the standard two-stage mediation framework. A mediation formula approach was proposed to estimate corresponding causal quantities using a piecewise linear approximation for the cumulative baseline hazard function of the interval-censored outcome. We conducted a simulation study to demonstrate low bias of mediation effect estimators for a wide range of complex hazard shapes. The method was applied to the Jackson Heart Study data to help illustrate the mechanisms of smoking induced interval-censored incident hypertension data.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2017 program
|
