Abstract 3509: Bleeding in Patients Undergoing Percutaneous Coronary Interventions: A Predictive Model From 302,152 Patients in the ACC-NCDR
Introduction Bleeding after percutaneous coronary intervention (PCI) is associated with an increased incidence of death and ischemic events. Being able to assess the risk of bleeding in patients needing revascularization would permit physicians to tailor therapy (e.g. sheath size, choice of anti-platelet/ anti-thrombin agents) to potentially mitigate the risk of peri-procedural bleeding.
Methods We analyzed data from the ACC-NCDR of 302,152 patients who had PCI from 1/1/2004 to 3/30/2006. A composite bleeding end-point was used that included entry site bleeding, retroperitoneal bleeding, gastrointestinal bleeding, genitourinary bleeding, and bleeding of unknown cause. Multivariable logistic regression was used to identify characteristics associated with bleeding and to develop a predictive model. Discrimination was assessed with the c-statistic and calibration with a calibration plot. The model was developed from randomly sampled 80% of the data set, and validated on the remaining 20%.
Results The frequency of composite bleeding was 2.5%. Factors independently associated with increased risk of bleeding included use of pre-procedural intraaortic balloon pump (odds ratio [OR] 1.93), cardiogenic shock (OR 1.77), female sex (OR 1.75), acute coronary syndrome (OR 1.42), advanced age (OR 1.35), chronic lung disease (OR 1.27), congestive heart failure (OR 1.18), cererbrovascular disease (OR 1.17), peripheral vascular disease (OR 1.15), hypertension (OR 1.15), and decreased glomerular filtration rate (OR 1.11). Factors associated with decreased risk included dyslipidemia (OR 0.90), prior PCI (OR 0.71), prior coronary artery bypass surgery (OR 0.92), and increased weight (per 5 kg increase) (OR 0.98).The c-statistic for the model was 0.73., and the model performed well in a calibration plot.
Conclusions Bleeding complications from PCI can be modestly well predicted with pre-procedural patient characteristics. Further work is required to determine the applicability of this model when different anti-thrombin and anti-platelet therapies are used. This model may identify patients at higher risk of bleeding associated with PCI and, thereby, help guide clinical decision making.