Objectives: Patients have few objective predictors of quality in their choice of physician. We studied whether physician education variables and board certification status were associated with 30 day mortality rates after percutaneous coronary interventions (PCI) in New York State.
Methods: Using the New York State PCI registry we obtained 30 day risk adjusted mortality rates after PCI procedures for all interventional cardiologists practicing between 2010-2012. Educational and certification variables were obtained using publicly available sources. Mortality rates were adjusted using a hierarchical Poisson shrinkage estimator. Hierarchical regression modeling was used to assess associations between mortality rates and education and certification markers (graduation from American vs. foreign medical school, years since medical school graduation, board certification in interventional cardiology and maintenance of certification for board certified physicians in 2016) with and without adjustment for caseload. Excluding correlated covariates, a regression model including caseload, board certification and US graduates was also constructed.
Results: A total of 346 interventional cardiologists performed an average of 427.41 ± 402.52 cases with 3.90 ± 3.51 deaths. The average shrinkage estimator adjusted mortality rate was 1.00 ± 0.16. Only interventional cardiology board certification (75.8% of providers) was associated with lower mortality rate in univariate analysis (β = -0.06; p = 0.03). No other variables had a significant association with the outcomes. After adjusting for caseload, the association with board certification became non-significant. In the multivariate regression model including the above noted covariates, the association between caseload and the outcome remained significant (β = -0.001; p <.01).
Conclusion: Risk-adjusted mortality rate after PCI is not associated with any education or certification markers, including board certification in interventional cardiology or maintenance of certification, after adjustment for caseload. Caseload was the only predictor of mortality rate in multivariate modeling.