Analytic Methods to Inform Interventions that Advance Cardiovascular Health Equity

PROJECT SUMMARY Racial disparities in health arise in complex ways. Multicomponent interventions to address them (often implemented by healthcare systems, employers, and community organizations) are often designed to address multiple mechanisms (e.g., medication adherence, lifestyle behaviors). However, not all such interventions are effective in reducing disparities in health. Designing effective interventions requires knowing which barriers to health, when acted upon, may greatly reduce disparity. Standard approaches (e.g., mediation analysis, Oaxaca- Blinder decomposition) are often used to generate this evidence base but do so under severe limitations, including their inattention to concepts of equity (e.g., they often over-adjust measures of racial disparity for factors implicated in disparity such as socioeconomic status), and their inattention to matters of causal inference (e.g., confounding). Causal decomposition methods, in contrast, answer the question: how much disparity in an outcome (e.g., hypertension control) would change if we could equitably remove racial disparity in a barrier to health (e.g., clinical inertia in treatment decision-making). They consider what variables should separately be accounted for when measuring disparity in an outcome, when removing disparity in a barrier, and while addressing confounding. This new approach is being taken up by applied researchers, but often they are interested in more general settings than current methods allow. The goals of the proposed methodological project are to extend causal decomposition methods to settings relevant for cardiovascular health equity research and provide guidance and tools to facilitate their adoption and use. This methodological project will extend and tailor equity-aligned causal decomposition methods to: 1) accommodate longitudinal and time-to-event data structures; 2) accommodate aspects of patient-centeredness and intersectionality; 3) accommodate generalization beyond the study sample to specific populations of interest. A key focus of the work will be to disseminate existing and newly developed causal decomposition methods to applied researchers in cardiovascular health equity through the development of tutorials, publicly available software tools, a website, and courses and workshops at research conferences. The methods development and dissemination will be informed by three illustrative data sets, the Coronary Artery Risk Development in Young Adults (CARDIA) study, the Reducing Inequities in Care of Hypertension: Lifestyle Improvement for Everyone (RICH LIFE) Project, and the All of Us Research Program (AURP). If this proposal is successful, the methods developed herein could support the design of effective, targeted, and multicomponent interventions that reduce disparity, e.g., those of the Health Equity Action Network and the RESTORE Health Equity Network.