Less Is More for Bleeding Management Algorithms in Cardiac Surgery: A Network Meta-Analysis and Meta-Regression of Randomized Studies.

Abstract

BACKGROUND Major bleeding after cardiac surgery is still a relatively common complication, requiring red blood cell (RBC) transfusions and use of procoagulants. The existing guidelines recommend a bleeding management based on viscoelastic tests and bleeding management algorithms (BMA). However, there are different BMAs with different trigger values prompting the use of different procoagulants; consequently, based on these trigger values, a BMA can be more or less liberal in the use of drugs and blood derivates aimed to control bleeding. At present, no studies have investigated the effectiveness of liberal versus restrictive BMAs in limiting the risk of RBC transfusion. METHODS In this study, we performed 2 network meta-analyses and a meta-regression of randomized and nonrandomized (before and after) studies where a BMA was applied. Based on the trigger values prompting the use of procoagulants (fresh frozen plasma, platelet concentrate, fibrinogen concentrate, prothrombin complex concentrate, recombinant activated factors), we have adjudicated the various BMAs to a restrictive or liberal group, and we compared the effectiveness of restrictive versus liberal BMAs in limiting RBC transfusions. Additionally, the consumption of procoagulants was compared between the 2 groups. RESULTS Both restrictive and liberal BMAs were superior to conventional strategies in limiting RBC transfusions (rate and units). Restrictive BMAs were more effective than liberal BMAs in terms of RBC units transfused (mean difference -0.43 units, 95% confidence interval [CI], -0.80 to -0.07, P = .020). The RBC transfusion rate was tested after correction for potential confounders (complexity of surgery and hematocrit trigger for RBC transfusion) with a meta-regression RBC transfusion rate was significantly lower in restrictive versus liberal BMAs (odds ratio 0.728, 95% CI 0.569-0.932, P = .012). The use of any kind of procoagulants was significantly (P = .001) lower in restrictive versus liberal BMAs. CONCLUSIONS Overall, viscoelastic test-based BMAs are superior to conventional strategies in limiting RBC transfusions in the presence of major bleeding; however, a restrictive strategy of procoagulant administration is superior to a liberal strategy in terms of RBC transfusion containment. Liberal BMAs are associated with a significantly higher use of procoagulants without any benefit in terms of RBC transfusions; therefore, in terms of cost/benefit ratio, restrictive BMAs should be preferred.