Chinese CABG Risk Assessment System

Objective:To construct a complex and a bedside mental arithmetic model for the prediction of in-hospital mortality in Chinese patients undergoing coronary artery bypass grafting (CABG).Methods:From2007to2008, information of9839consecutive CABG patients was collected in Chinese Coronary Artery Bypass Grafting Registry which recruited43Chinese centers. This database was randomly divided into developmental and validation subsets (9:1). A complex risk model and a simple score were developed using logistic regression with the primary endpoint of in-hospital mortality. The simple ABE score was computed as:age(years)*age(years)/(body weight(kg)*EF(%)).A cutoff point was chosen to achieve mental arithmetical convenience. Calibration and discrimination characteristics were assessed in the validation dataset. Thresholds were defined for each model to distinguish different risk groups. Both models were compared with EuroSCORE and ACEF score in the validation dataset. Postoperative events, including postoperative morbidity (postoperative stroke, renal failure, deep sternal wound infection, reoperation for any reason, and composite morbidity), postoperative ventilation time, postoperative ICU stay, postoperative length of stay and total costs, were further compared between high risk and low risk groups defined by cutoff points of both complex and simplified models.Results:In the developmental dataset, calibration by Hosmer-Lemeshow (HL) test were p=0.44and p=0.41, discrimination by area under ROC (AUC) were0.80and0.74for complex and simple model. In the validation dataset, HL test were p=0.34and p=0.86, AUC were0.78and0.77, respectively. The performance for both models turned out good. Similar or better accuracy and calibration were found for both models compared with EuroSCORE (HL p=0.60; AUC0.73) and ACEF score (elective procedures only, HL p=0.52; AUC=0.69). A cutoff point of ABE=1was set and ABE≥1was a risk factor for mortality and all the other events. Similarly, a cutoff point of4was set for the complex model and a score≥4was proved to be a risk factor for all the postoperative events.Conclusion:An additive complex model and a simplified ABE score were developed based on most up-to-date data in China and both showed similar or better accuracy compared with EuroSCORE in this set of Chinese patients. With a cutoff point of ABE=1, ABE score can be used conveniently at bedside using mental arithmetic. Moreover, both models have predictive value for postoperative morbidity, prolonged ventilation time, prolonged postoperative ICU stay, prolonged length of stay as well as increased total costs.