The AusScore—A Validated Model for Predicting Outcome Following Isolated Coronary Bypass Graft Procedures in Australia

Christopher M. Reid, Baki Billah, Diem Dinh, Julian A Smith, Gilbert Shardey, on behalf of the ASCTS Database Group

Research output: Contribution to journalMeeting Abstractpeer-review

Abstract

Background: Recently, it was shown that the widely used EuroSCORE system validated poorly for an Australian Cohort hence the need for a valid model is required to appropriately guide surgeons and patients in assessing CABG preoperative risk. Methods: Seven thousand seven hundred and nine patients undergoing isolated CABG surgery between July 2001 and June 2005 in Victorian Public hospitals were included for analysis. Data were divided into two sets for model creation (66%) and validation (34%). Boot­strap and backward elimination methods (1000 repeated samples) were used to identify variables that are inde­pendent predictors of mortality which were entered into multiple logistic regression analysis to develop a preoper­ative predictive model. The predictive AusSCORE model was assessed using the validation data set for receiver operating curve characteristics (ROC) and p-values for Hosmer-Lemeshow goodness-of-fit test. Results: Factors selected as independent predictors were: age group, NYHA class II or IV, ejection fraction, cerebrovascular disease, urgency of procedures, previ­ous CABG, hypercholesterolaemia, peripheral vascular disease, cardiogenic shock, sex, smoking status and medications: inotropes. The area under the receiver operator curve was 0.8219 and the p-value for the Hosmer-Lemeshow chi-squared test statistic is 0.2404 which indicates the model fit the data well. Conclusion: We have developed a preoperative 30-day mortality risk prediction model for Isolated CABG pro­cedures and recommend its use in the Australian setting.
Original languageEnglish
Article number442
Number of pages1
JournalHeart Lung and Circulation
Volume16
Issue numberSuppl. 2
DOIs
Publication statusPublished - 2007

Cite this