Hospital and Physician Variability in Revascularization Decisions and Outcomes for Patients With 3-Vessel and Left Main Coronary Artery Disease: A Population-Based Cohort Study

Background Hospital- and physician-level variation for selection of percutaneous coronary intervention versus coronary artery bypass grafting (CABG) for patients with coronary artery disease has been associated with outcome differences. However, most studies excluded patients treated medically.Methods and Results From 2010 to 2019, adults with 3-vessel or left main coronary artery disease at 3 hospitals (A, B, C) in Alberta, Canada, were categorized by treatment with medical therapy, percutaneous coronary intervention, or CABG. Multilevel regression models determined the proportion of variation in treatment attributable to patient, physician, and hospital factors, and survival models assessed outcomes including death and major adverse cardiovascular events over 5 years. Of 22 580 patients (mean age, 67 years; 80% men): 6677 (29%) received medical management, 9171 (41%) percutaneous coronary intervention, and 6732 (30%) CABG. Hospital factors accounted for 10.8% of treatment variation. In adjusted models (site A as reference), patients at sites B and C had 49% (95% CI, 44%-53%) and 43% (95% CI, 37%-49%) lower rates of medical therapy, respectively, and 31% (95% CI, 24%-38%) and 32% (95% CI, 24%-40%) lower rates of CABG. During 5.0 years median follow-up, 3287 (14.6%) patients died, with no intersite mortality differences. There were no between-site differences in acute coronary syndromes or stroke; patients at sites B and C had 24% lower risk (95% CI, 13%-34% and 11%-35%, respectively) of heart failure hospitalization.Conclusions Hospital-level variation in selection of percutaneous coronary intervention, CABG, or medical therapy for patients with complex coronary artery disease was not associated with differences in 5-year mortality rates. Research and quality improvement initiatives comparing revascularization practices should include medically managed patients.