Prediction of risk for bleeding, myocardial infarction and mortality after percutaneous coronary intervention in patients with acute coronary syndromes

BACKGROUND: Whether bleeding and myocardial infarction (MI) improve the performance of risk prediction models for mortality in patients with acute coronary syndromes (ACS) treated with percutaneous coronary intervention (PCI) remains unknown. METHODS: This study included 3377 patients with ACS who underwent PCI in the setting of the ISAR-REACT 5 trial. Patients with bleeding, MI or those dying at 1 year after PCI were characterized in terms of baseline characteristics, risk estimates and C-statistic of the risk prediction models for these outcomes. RESULTS: Major bleeding (Bleeding Academic Research Consortium types 3-5), MI and mortality occurred in 195 patients (5.8%), 143 patients (4.3%) and 143 patients (4.3%), respectively. After adjustment, bleeding [hazard ratio = 5.08; 95% confidence interval (CI), 3.03-8.53; P < 0.001] and MI [hazard ratio = 5.90; 95% CI, (3.00-11.65); P < 0.001) remained independently associated with the risk for 1-year mortality. The C-statistic (with 95% CI) of the model for bleeding, MI and mortality was, 0.755 (0.722-0.786), 0.752 (0.717-0.789) and 0.868 (0.837-0.896), respectively. The inclusion of bleeding [C-statistic: 0.892 (0.867-0.913); delta C-statistic 0.024 (-0.014 to 0.065); P = 0.200] or MI [C-statistic: 0.878 (0.851-0.903); delta C-statistic 0.011 (-0.030 to 0.053); P = 0.635] in the risk prediction models for mortality alongside baseline demographical and clinical variables did not improve prediction for mortality. CONCLUSIONS: In patients with ACS treated with PCI, mortality is the most accurately predicted outcome. Bleeding and MI did not improve risk discrimination for mortality when added in the risk prediction models for mortality suggesting that these outcomes do not provide incremental prognostic information on top of baseline demographical and clinical data