Abstract 20862: Bayesian Number-Needed-to-Treat Analysis of Rosuvastatin for Prevention of Venous Thromboembolism in Patients with Elevated High Sensitivity C-Reactive Protein
Background: Recent clinical trials have investigated the role of statin therapy on the prevention of thromboembolic events. The JUPITER trial specifically evaluated a subset of patients particularly susceptible to the benefits of rosuvastatin. However, controversy exists regarding the benefit of treating individuals with intermediate risk. We sought to explore this controversy with using Bayesian analytic techniques, including a Bayesian form of NNT analysis.
Methods: Bayesian posterior probabilities (Pr), in relation to probability of benefit versus harm, as well as probabilities of greater than 20% benefit (Pr>20%), were calculated. Subsequently, a graphical sensitivity analysis was performed with each endpoint. The analysis was conducted with experiential probabilities obtained from the JUPITER trial data, and a null prior probability with a mean RR of 1, and a standard deviation of 2. Bayesian analysis was performed using two methods, with MCMC simulation. Bayesian NNT was estimated using established methods.
Results: Bayesian analysis of the total venous thromboembolism (VTE) events indicated a Pr of 99.5% benefit and a Pr>20% of 95.8%. This indicates a clinically relevant benefit (generally thought of as higher than a Pr>20% greater than 80%). DVT only benefits were Pr of 96.7% and a Pr>20% of 95.1%. This trend continued in patients with VTE, but no prior cardiac events.
Results for pulmonary embolism, death after VTE and VTE after cardiac events did not meet the threshold for clinical relevance. Bayesian NNT point estimate for total VTE is 10.86.
Conclusions: The Bayesian analysis suggests a clinically important benefit of rosuvastatin in the prevention of total VTE events in patients with increased values of high sensitivity C-reactive protein. Bayesian NNT values strongly support the use of statin therapy in low-intermediate risk patients. Further, NNT values improve upon conventional estimates reported in the JUPITER study (NNT of 25) and subsequent analysis using Kaplan-Meier curves. The intervention did not have a clinically important effect in patients with VTE after having cardiac events.
- © 2010 by American Heart Association, Inc.