Primary Submission Category: Matching, Weighting
Achieving Covariate Balance in Infant RSV Prevention through Cardinality Matching with Multiple Treatment Options
Authors: Lauren Liao, Karen Jacobson, Andrew Watson, Sally Stephens, Nicola Klein, Samuel Pimentel,
Presenting Author: Lauren Liao*
Treatment variation, such as different treatment options, often occurs in practice. A study that primarily aims to compare treatment versus control may also examine the effectiveness of different treatment options. For example, when comparing individuals receiving any treatment versus none, the study may initially focus on the overall treatment effect, ignoring variation among treatment options, while later analyses examine the effects of each option separately. Traditional matched designs targeting the overall comparison (any treatment vs. none) may fail to guarantee balanced comparisons for additional analyses involving different treatment options. We propose a single matched study design that ensures balanced comparisons for the overall comparison and separate treatment options between the treated and control subjects. We leverage and extend the cardinality matching approach to create covariate balance constraints for overall and separate treatment option comparisons and impose additional constraints to ensure each individual can only receive one treatment option. We demonstrate this method in a study of newborn infants at risk for respiratory syncytial virus (RSV), evaluating both the overall effect of receiving any RSV protection and the effects of distinct treatment options for protection (maternal vaccination or infant monoclonal antibody treatment).