An Integrated Approach to Assess Heterogeneity in Health Disparity

• Activity Number: 314 - Pioneering Statistical Methods to Alleviate Health Disparity and Achieve Health Equity
• Type: Invited
• Date/Time: Tuesday, August 9, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Chen-Pin Wang* and Booil Jo
• Companies: University of Texas Health San Antonio and Stanford University
• Keywords: latent variable; statistical learning; disparity; Oaxaca-Blinder

Abstract

We proposed an integrated approach under the potential outcome framework to assess the heterogeneity in health disparity that requires between group comparisons to be adjusted for multiple intermediate variables over time. Built on our prior work that integrated general latent variable modeling with targeted statistical learning, we derived the trajectory strata associated with the intermediate variables under the inverse propensity scores weighted sample. Next, we derived stratum-specific disparity (i.e., principal effects) using the Oaxaca-Blinder estimator. To overcome model non-identifiability, we assumed invariant prediction and superpopulation in place of the monotonicity and exclusion restriction assumptions. Finally, we conducted sensitivity analyses to assess the robustness of the proposed method using two types of sensitivity bounds jointly– the differences in disparity across trajectory strata, and the joint distribution of the trajectory strata under each comparison group.