AusSCORE II in Predicting 30-Day Mortality Following Isolated CABG in Australia

Baki Billah, Molla Huq, Gilbert Shardey, Julian A Smith, Andrew E. Newcomb, Fahim Sufi, Christopher M. Reid, Dhenisha Dahya

Research output: Contribution to journalMeeting Abstractpeer-review

Abstract

Introduction: The Australian System for Cardiac Operative Risk Evaluation (AusSCORE) is a 30-day mortality risk prediction model which was developed for predicting 30-day mortality following isolated CABG in the Australian population (1). The AusSCORE model was published in 2009 utilising data from the Australian and New Zealand Society of Cardiac and Thoracic Surgeons (ANZSCTS) Database. Since then the database has experienced significant expansion. The AusSCORE requires updating to account for this expansion and take advantage of modernised risk model development techniques. Thus the objectives of this study are to: (i) update the original AusSCORE with a larger contemporary dataset, (ii) develop and validate the model using bootstrap methods rather than split-sample methods (2), (iv) evaluate model specification and (v) develop an online calculator for 30-day mortality risk prediction. Methods: Data were collected by the ANZSCTS Database from 25 hospitals across Australia from 2001 to 2011. A total of 31,250 patients had isolated CABG and the primary outcome was 30-day mortality. Bootstrapping methods were used to construct the updated logistic model (AusSCORE II). Model specification was evaluated using multiple sets of 1000 bootstraps. Themodel’s prediction performance was evaluated using bootstrap validation methods instead of split sample method. A risk algorithm was derived to stratify patients as being at low, moderate or high risk. An online calculator was developed to predict preoperative risk of 30-day mortality following isolated CABG. The prediction performance of the original AusSCORE and AusSCORE II was compared. Results: From 2001 to 2011 31,250 patients had CABG in Australia. The average age was 65.6 12.9 years and 78.6% were male. The following 13 variables were selected as independent risk factors for 30-day mortality following CABG: age, gender, ejection fraction estimate, New York Heart Association (NYHA) class, inotropic medication, chronic kidney disease (CKD) stages, and urgency of procedures, nitrate medication, previous MI, anticoagulant medication, previous cardiac surgery and cardiogenic shock. The bias corrected discrimination and calibration of the model were very good (discrimination ROC: 82.0%, calibration slope: 0.987). The overall observed and predicted mortality were similar (observed: 1.63, predicted: 1.57). The observed versus predicted 30-day mortality in the low, moderate and high risk groups were respectively 0.4% versus 0.5%, 1.9% versus 1.8% and 8.4% versus 8.9%. Discussion: AusSCOREII is an important tool for cardiac surgeons allowing validation of clinical decision making and advancing strategy to improve 30-day mortality. It is also an important tool for referring physicians, improving the patient consultation.
Original languageEnglish
Pages (from-to)e48–e49
Number of pages2
JournalHeart Lung and Circulation
Volume23
Issue number1
DOIs
Publication statusPublished - Jan 2014

Cite this