Introduction:
Calculating the net benefit between bleeding risks and ischemic complications of antithrombotic therapy remains challenging.
Hypothesis:
Hierarchical meta-analyses may allow inferences to be made about the parameter θ, which reflects the underlying hypothesis that one therapy is superior to another conditional on the types of risks associated with various end points in clinical trials.
Methods:
We employed Bayesian hierarchical meta-analyses of 10 randomized clinical trials randomizing 34,658 patients to either bivalirudin or heparin, with or without platelet glycoprotein inhibitors, during percutaneous coronary intervention (PCI).
Results:
No difference was seen in the rate of death 30 days after using bivalirudin or heparin (OR 1.01, 95% Bayesian credible interval [BCI] 0.79-1.28), but the risk of major bleeding was lower (OR 0.58, 95% BCI 0.38-0.84) and the risk of definite stent thrombosis (ST) was higher (OR 1.86, 95% BCI 1.19-2.94) after using bivalirudin than after using heparin. For every 100 patients, using bivalirudin in place of heparin prevented 4.1 episodes of major bleeding episodes but caused 0.8 episode of definite ST. (Figure Legend: Probability density functions normalized to 1 for risk of major bleeding, death and definite stent thrombosis at 30 days after treatment with bivalirudin or heparin, with or without platelet glycoprotein inhibitors.)
Conclusions:
Neither bivalirudin nor heparin emerges as a preferred strategy in an analysis that accounts for mortality differences. Because bivalirudin reduces major bleeding at the cost of increased stent thrombosis, decisions between bivalirudin and heparin based on baseline bleeding risk and ST risk are likely to maximize net clinical benefit.