Primary Submission Category: Mediation
Mediation Analysis for Case-Control Studies with Secondary and Tertiary Outcomes
Authors: Zichun Cao, Honglei Chen, Chenxi Li, Yuehua Cui, Aimee D’Aloisio, Dale Sandler, Zhehui Luo,
Presenting Author: Zichun Cao*
Background: A case-control study is an epidemiological study design usually for studying the association of risk factors for a rare disease. It is not uncommon that researchers use the case-control sample for additional secondary or tertiary outcomes of interest. Statistical methods for association analysis using such secondary or tertiary outcomes have been proposed but the literature is limited for mediation analyses using such data. With the increasing amount of secondary or tertiary data, plus the fact the prospective cohort study and randomized controlled trials are expensive for mediation analysis, there is a need for using small-scale and less costly data to do preliminary mediation analysis.
Goal: To perform mediation analysis for a target population (e.g., a cohort) using case-control samples with additionally collected variables.
Methods: We proposed a new weighting method for estimating mediating effects (controlled direct effects and natural direct and indirect effects) using case-control data with a further collected (secondary) variable as the mediator and another tertiary variable collected even later as the outcome. And we also implemented the proposed estimation method in an empirical application using the NIEHS Sister Study for illustration.
Results: The proposed estimators performed very well in the Monte Carlo simulations. The implementation of the method is easy and smooth.