Abstract 3441: Bedside Estimation of Risk From Percutaneous Coronary Interventions: The New Mayo Clinic Risk Scores?
Background. Existing risk-adjustment models for percutaneous coronary interventions (PCI) include angiographic variables, thus limiting the pre-procedure risk assessment and counseling of patients. We proposed to build risk models for PCI outcomes from simple clinical and laboratory variables for pre-procedure risk stratification.
Methods and Results. Using the Mayo Clinic registry (January 2000 through April 2005), there were 7,547 initial PCI procedures conducted at Mayo Clinic and used for these analyses. Logistic regression was used to model the calculated risk score and major procedural complications. Separate risk models were made for mortality and major adverse cardiovascular complications (MACE). The final risk scores for both in-hospital mortality and MACE contained the same seven variables (age, MI<24 hours, shock, serum creatinine, ejection fraction, congestive heart failure, and peripheral vascular disease). The models had adequate goodness of fit and the areas under the receiver operator curve (ROC) were 0.76 and 0.89 for MACE and in-hospital mortality, indicating excellent overall discrimination. The prediction model was robust across many subgroups, including elective PCI, patients with diabetes, and elderly. Bootstrap analysis indicated that the model was not overfit to the modeling data set.
Conclusions. Seven variables can be combined into a convenient risk scoring system that accurately predicts cardiovascular complications after PCI. By not requiring the use of subjective and angiographic variables, these prediction models can be widely used by health care providers for pre-procedure risk stratification and counseling of patients.