Abstract Body (Do not enter title and authors here): Background: Diabetes affects 14.7% of adults in the US, contributing to health and economic burdens, with persistent differences in prevalence by race and ethnicity. Prior research suggests physician and patient engagement is vital to diabetes management.

Research Question: We investigated how clinical, behavioral, and sociodemographic context and characteristics affecting the patient-physician relationship were associated with diabetes HEDIS quality-of-care measures.

Methods: We performed a cross-sectional study of patients with commercial or Medicare Advantage coverage and their attributed physician using 2020-2023 data from one large, national payer. The primary outcomes were patient performance on four HEDIS measures: 1) dispensed ≥ one statin, 2) ≥80% adherence to statin, 3) retinal eye exam, and 4) kidney health evaluation. We used a linear probability model with physician fixed effects (FE) to explore the impact of patient and area-level social metrics on diabetes care quality, including sociodemographics; health behaviors, status, and care utilization; patient-physician social factor concordance; and patient-area-level factors. The physician FE approach controls for all physician-level confounders, such as practice location and prescribing tendencies.

Results: The samples included 63,205-100,827 physicians and 598,318-1,342,393 patient-years. The patients were 65.3% NH white, 15.6% NH Black, 13.2% Hispanic, and 5.9% NH Asian. A majority were age 45-64 (68.8%), male (53.0%), and had commercial coverage (84.8%). Performance on HEDIS measures was 67.6% (statin receipt), 76.9% (statin adherence), 51.1% (eye exam), and 48.7% (kidney evaluation), with inconsistent variation by race and ethnicity. In models with physician FE, age, number of primary care visits, annual wellness exam, and number of comorbidities consistently showed positive and significant associations with the outcomes. Compared to patients with no visit, having an annual wellness exam increased HEDIS performance by 1.6-7.6 percentage points across all outcomes. Other factors showed inconsistent relationships across outcomes.

Conclusion: Clinical factors such as frequent care utilization correlate with performance on diabetes care measures related to early detection or complication risk reduction. This modest differential in performance can translate into substantial population-level impacts. Efforts to improve diabetes care should prioritize consistent patient-physician engagement.