Meaningful causal decompositions in health equity research: definition, identification, and estimation through a weighting framework
share
Causal decomposition analyses can contribute to the evidence base for interventions that address health inequities. Through study design and assumptions, they rule out alternate explanations such as confounding, selection-bias, and measurement error. However, their practical use is impeded by critical challenges. First, current approaches pay little attention to what variables a disparity measure itself should condition on or standardize over. Second, current approaches have ignored what hypothetical interventions should condition on and thus may not reflect equity concerns that actual interventionists would respect. Third, there are several estimators across the epidemiology and economics literature but their incorporation of equity value judgements has not been examined, and these do not cover many of the ways in which disparities and hypothetical interventions may be defined. In this paper, motivated by the clinical example of treatment intensification and hypertension control disparities, we address these issues of defining and estimating meaningful causal decompositions. We present a framework that explicitly considers what the disparity measure and hypothetical intervention account for and how these choices can be mapped to notions of equity. For these general estimands we provide identifying assumptions and estimators based on adaptations of ratio-of-mediator probability and inverse-odds-ratio weighting and evaluate their statistical performance.
READ FULL TEXT