Coronary Assessment in Heart Failure within a Safety-Net Setting: Disparities and Outcomes.

Background: Though ischemic cardiomyopathy is the leading cause of heart failure (HF), most patients do not undergo coronary assessment after heart failure diagnosis. In a safety-net population, referral patterns have not been studied, and it is unknown whether coronary assessment is associated with improved HF outcomes. Methods: Using an electronic health record cohort of all individuals with HF within San Francisco Health Network from 2001-2019, we identified factors associated with completion of coronary assessment (invasive coronary angiography, nuclear stress, or coronary computed tomographic angiography). Then we emulated a randomized clinical trial of elective coronary assessment with outcomes of all-cause mortality and a composite outcome of mortality and emergent angiography. We used propensity scores to account for differences between groups. We used national death records to improve ascertainment of mortality. Results: Among 14,829 individuals with HF (median 62 years old, 5,855 [40%] women), 3,987 (26.9%) ever completed coronary assessment, with 2,467 (18.5%) assessed out of 13,301 with unknown CAD status at HF diagnosis. Women and older individuals were less likely to complete coronary assessment, with differences by race/ethnicity, medical history, substance use, housing, and echocardiographic findings. Among 5,972 eligible for inclusion in the "target trial," 627 underwent early elective coronary assessment and 5,345 did not. Coronary assessment was associated with lower mortality (HR 0.84; 95% CI 0.72-0.97; p=0.025), reduced risk of the composite outcome, higher rates of revascularization, and higher use of medical therapy. Conclusions: In a safety-net population, disparities in coronary assessment after HF diagnosis are not fully explained by CAD risk factors. Our target trial emulation suggests coronary assessment is associated with improved HF outcomes possibly related to higher rates of revascularization and GDMT use, but with low certainty that this is finding is not attributable to unmeasured confounding.