Predictors of excessive blood use after coronary artery bypass grafting. A multivariate analysis

One hundred fifty-nine consecutive patients who underwent artery bypass grafting were studied to determine clinical and laboratory predictors of excessive postoperative packed red blood cell transfusion. Consideration of the distribution of packed red blood cells administered revealed that the patients could be divided into two groups: those patients who received 5 units of red blood cells or less (group I, n = 139) and those patients who received more than 5 units of packed red blood cells (group II, n = 20). The Mann-Whitney test or Fisher's exact test was used whenever appropriate to test differences between these two groups with respect to twelve patient variables. Patients in group II were found to have a significantly longer preoperative template bleeding time and decreased preoperative packed red blood cell volume (p < 0.0008 for both variables). In addition, group II patients were significantly older (p = 0.026), were more likely to have had preoperative heparin therapy (p = 0.049), and contained a greater proportion of women (p = 0.0048). Of interest, variables that did not achieve statistical significance between groups were partial thromboplastin time, prothrombin time, platelet count, preoperative hematocrit level, urgency of operation, recent ingestion of aspirin, and recent heparin administration. All of the measured variables were used in a stepwise logistic regression analysis to identify the best predictors of the need for more than 5 units of packed red blood cells after operation. Of the variables examined, bleeding time (p < 0.001; χ² improvement = 15.1) and red blood cell volume (p = 0.009; χ² improvement = 6.8) were the best predictors of excessive postoperative packed red blood cell use. On the basis of a 50% logistic probability level, the specificity and sensitivity of these two variables in predicting greater than a 5-unit tranfusion requirement were 85% and 99%, respectively. A clinically useful nomogram based on this logistic model is presented. This nomogram suggests that a ratio of bleeding time to red blood cell volume of 0.0071 or greater is associated with a greater than 70% chance of requiring more than 5 units of packed red blood cells. We conclude that preoperative bleeding time and red blood cell volume are useful predictors of excessive postoperative blood transfusion. These results suggest that factors other than aspirin therapy may be associated with bleeding time prolongation leading to excessive postoperative transfusion.