Activity Number:
|
258
- SPEED: Causal Inference and Related Methodology
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 2:00 PM to 2:45 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #332614
|
|
Title:
|
Balancing Scores Weighing Methods and Sensitivity Analysis to Unfold Health Disparity
|
Author(s):
|
Chen-Pin Wang*
|
Companies:
|
University of Texas Health San Antonio
|
Keywords:
|
health disparity;
balancing score;
weighting;
rank-and-replace;
National Academy of Medicine;
sensitivity analysis
|
Abstract:
|
The National Academy of Medicine (NAM) defines the disparity in a health outcome as the between group difference in the health outcome that is attributed to the "not allowed to be adjusted" covariates (Xn) conditioned on the "allowed to be adjusted" covariates (Xa) being equalized between groups. Two types of NAM-concordant disparity measures developed under the potential outcome framework are Rank-and-Replace (R-R) and balancing scores weighing (BSW; Dawid 1979). These methods may not yield a realistic estimate of health disparity given the uncertainty of (i) the correlation between Xn and the adjusted Xa and (ii) the predictivity of Xn conditioned on the adjusted Xa. This paper proposes (1) a framework that integrates BSW with R-R to decompose the overall disparity into the disparity due to the between group difference in the predictivity of Xn and the difference in the distribution of Xn; and (2) a sensitivity analysis scheme to project the range of disparity under plausible Cor(Xa,Xn) and predictivity of Xn. Statistical assumptions and implications to feasible interventions for reducing disparity will be discuss. An application to cancer disparity will be demonstrated.
|
Authors who are presenting talks have a * after their name.