Abstract 3917: The Toronto Risk Score for Major Adverse Cardiovascular Events Following Percutaneous Coronary Intervention
Objective: To develop and internally validate an objective, contemporary risk index for major adverse cardiovascular events (MACE) following percutaneous coronary intervention (PCI) which facilitates the identification of high risk patients and the calculation of risk-stratified outcomes.
Methods: Logistic regression for the binomial outcome MACE (defined as death, MI, coronary bypass surgery (CABG), or stroke) was performed in 9803 patients undergoing PCI between April 2000 and April 2005. Odds ratios for the independent predictors of MACE were rounded to their nearest integer to provide a risk weight for each variable. Risk weights which characterized each patient were summed to create the Toronto PCI Risk Score. Weighted linear regression was used to evaluate the calibration curve of predicted versus observed MACE for each unit of the risk score. Internal validation was performed by bootstrap resampling. Quartiles of the risk score were used to construct relative risk groups: Low (23%), Moderate (29%), Moderate-High (26%), High (22%).
Results: MACE occurred in 3.5% of patients. Risk weights derived from the rounded odds ratios are contained in parentheses: age per decade over 49 (1), female (2), NYHA 4 (2), LVEF<20% (2), MI within 1 yr of PCI (2), multivessel disease (2), ACC type C as worst type of target lesion (2), renal insufficiency, defined as serum creatinine>133 mmol/L or dialysis (2), carotid disease (2), thrombolysis (2), evolving MI (3), shock (7). The model had a ROC of 0.74, Hosmer-Lemeshow goodness of fit p=0.7, and excellent calibration (r2 = 0.96, p<0.001). All score variables occurred in >50% of the bootstrap models. Risk-stratified outcomes are presented in the Table⇓. All P values were <0.001.
Conclusions: The Toronto PCI Risk Score is a reliable measure of acuity which can identify patients who are at high risk of experiencing a major adverse cardiovascular event following PCI, and permit evaluation of risk-stratified outcomes.