ABSTRACT Health-related social needs, particularly food insecurity, housing instability, and transportation barriers, are associated with poor outcomes for people with type 2 diabetes mellitus (T2DM). In particular, these factors are associated with worse glycemic, blood pressure, and LDL cholesterol control, which significantly increases the risk of macrovascular and microvascular complications. They are also key drivers of racial/ethnic and socioeconomic status-based disparities in diabetes outcomes. Thus, there is a growing call for interventions to address these needs and improve T2DM outcomes. However little research has assessed whether changes in health-related social needs are associated with changes in T2DM outcomes. This knowledge gap hampers efforts to develop effective interventions. Addressing it will provide much- needed evidence on which patients to screen for which social needs, and on which interventions targeting social needs are most likely to improve T2DM outcomes. The pathways linking social needs and T2DM outcomes are likely characterized by interactions between individual- (e.g., needs, age, gender, race/ethnicity), clinic- (e.g., clinic characteristics; specifics of the intervention), and area-level factors (e.g., local economic conditions; community resources). Given this complexity, a multi-pronged approach is needed. We will combine 3 methods of investigation: 1) longitudinal, multi-level, regression analysis; 2) innovative machine learning methods to detect novel combinations of factors associated with heterogeneous response to health- related social needs and health-related social needs interventions while avoiding spurious findings; and 3) thoughtful qualitative investigation. Such an analysis has never before been possible, because the needed data elements have not been united. This proposal seeks to answer whether improvements in specific health-related social needs are associated with improvements in specific clinical outcomes, in what circumstances, and which approaches to addressing health-related social needs, if any, best improve outcomes. We will leverage what we believe to be the nation’s largest dataset of patient-reported health-related social needs, clinical outcomes, and community- and clinic-level data. We will examine whether changes in health-related social needs are associated with changes in hemoglobin A1c, systolic and diastolic blood pressure, and LDL cholesterol. Further, we will evaluate whether clinic-based interventions seem to improve these outcomes, and if there are important variations in these interventions that are associated with different response to the intervention. The proposed work will yield important, previously unavailable evidence on how to refine and improve health-related social need interventions. Specifically, it will help us understand better how better to care for a population at high risk for T2DM complications. Overall, this project will substantially advance NIDDK’s ...