Abstract 2490: The Nova Scotia Non-ST Elevation Acute Coronary Syndrome Mortality Risk Score
Introduction: Existing acute coronary syndrome (ACS) risk scores generally predict short term events. We have developed a non-ST elevation ACS admission risk score to predict one year mortality in a population-based Canadian setting.
Methods: We identified 14142 patients hospitalized with unstable angina or non-ST elevation MI (NSTEMI) between 1998 and 2002 using the Improving Cardiovascular Outcomes in Nova Scotia (ICONS) database. Patients with shock, cardiac arrest or death within 24 hrs were excluded. Using a multivariate discrimination model, independent predictors of one year mortality were entered into logistic regression to calculate a risk score. The model was internally validated and its predictive accuracy assessed using receiver operator characteristic analyses and C-statistics.
Results: Ten variables with no interaction independently predicted mortality. An adjusted score was assigned to each variable based on its regression coefficient. Age >70, NSTEMI, prior CHF, prior renal failure or creatinine >133 μmol/L each = 3 points; prior stroke, atrial fibrillation, hemoglobin <100 g/L, diastolic BP < 50 mmHg or heart rate >100/min each = 2 points; prior MI, diabetes or random glucose >11 mmol/L each = 1 point. Patients were stratified into five groups based on one year mortality (p<0.001; fig 1⇓). The predictive accuracy (C-statistic 0.83 ± 0.01) was significantly greater than for other published risk scores.
Conclusions: The Nova Scotia NSTEACS risk score is highly predictive of one year mortality. We anticipate that more reliable mortality prediction will facilitate clinical decision making as well as further study of treatment-risk interactions.