Predictive model for blood product use in coronary artery bypass grafting

Objective: To build a clinical predictive model to determine the need for transfusing blood and its products in coronary artery bypass grafting (CABG) procedures in South East Asian population. Study Design: Analytical study. Place and Duration of Study: Section of Cardiothoracic Surgery, Aga Khan University Hospital, Karachi, from January 2006 to October 2014. Methodology: Information on pre-, intra- and postoperative variables were collected for all adult patients who underwent on-pump CABG. The patients grouped into those who received blood and its components, and those who did not. A univariate as well as multivariate logistic model was built to determine the predictors of transfusion. Results: A total of 3,550 patients underwent CABG and males were dominant in both groups (75 vs. 93%). The transfusion rate was 56.4% (n=2001). Age (adjusted OR 1.03, p < 0.001), obesity (1.50, p=0.001), tobacco use (1.29, p=0.001), and male gender (4.51, p < 0.001) found to be a stronger predictor. Among preoperative comorbidities, diabetes (1.20, p=0.016), myocardial infarction (1.22, p=0.009), preoperative creatinine (1.12, p=0.033), and left main vessel disease of > 50% (1.49, p < 0.001) were independently associated with the outcome. Compared to elective cases, transfusion rates were high in urgent and emergent cases (OR: 1.93 and 3.36 respectively, p < 0.001 for both). Conclusion: Age, male gender, obesity, tobacco use, diabetes, myocardial infarction, high creatinine, urgent and emergent cases were independent predictors of transfusion in CABG procedure. This model can be utilized for preoperative risk stratification of patients and their management to improve the outcomes.