Background: Previous data on bleeding after percutaneous coronary intervention (PCI) have been obtained primarily from randomized trials that focused on in-hospital bleeding. The incidence of late bleeding after PCI, its independent predictors, and its prognostic importance in clinical practice has not been fully addressed.
Methods and results: We evaluated 22 798 patients aged >65 years who underwent PCI from December 1, 2003, to March 31, 2007, in Ontario, Canada. Cox proportional hazard models were used to determine factors associated with late bleeding, which was defined as hospitalization for bleeding after discharge from the index PCI, and to estimate risk of death or myocardial infarction associated with late bleeding. We found that 2.5% of patients were hospitalized for bleeding in the year after PCI, with 56% of bleeding episodes due to gastrointestinal bleed. The most significant predictor of late bleeding was warfarin use after PCI (hazard ratio [HR], 3.12). Other significant predictors included age (HR, 1.41 per 10 years), male sex (HR, 1.24), cancer (HR, 1.80), previous bleeding (HR, 2.42), chronic kidney disease (HR, 1.93), and nonsteroidal antiinflammatory drug use (HR, 1.73). After adjusting for baseline covariates, hospitalization for a bleeding episode was associated with a significantly increased 1-year hazard of death or myocardial infarction (HR, 2.39; 95% CI, 1.93 to 2.97) and death (HR, 3.38; 95% CI, 2.60 to 4.40).
Conclusions: Hospitalization for late bleeding after PCI is associated with substantially increased risk of death and myocardial infarction. The use of triple therapy (i.e., aspirin, thienopyridine, and warfarin) is associated with the highest risk of late bleeding.