Predictive models of prolonged stay after Coronary Artery Bypass surgery

The prolonged hospitalisation after cardiac surgery can significantly describe the patients' health status. Many studies have assessed the risk factors associated with prolonged stay after cardiac surgery in hospital but only a few predictive models have been developed. Therefore, this study is aimed to build a logistic regression and decision tree model in predicting the prolonged stay for Coronary Artery Bypass Graft surgery (CABG) patients after surgery. This study included 3560 patients who received CABG from a heart surgery centre in Malaysia for a period of three years (20072010). Risk factors such as age, diabetes, hypertension, gender, obesity, chest re-operation and complications (such as stroke, wound infection and atrial fibrillation) after surgery were determined by the predictive models. Overall, 2499 of CABG patients were discharged within less than 10 days, whereas 1061 patients required prolonged ≥10 days stays. The study population consisted of patients aged 21 to 85 years with 13.1% of female and 76.9% of male patients. From the analysis, predictive accuracy for decision tree (CHAID) and logistic regression (Enter) model were 65.86% and 75.87% respectively. Logistic model revealed that age, diabetes mellitus, chest re-operation (I), atrial fibrillation and wound infection influenced the CABG patients' prolonged length of stay.