DESCRIPTION (provided by applicant): Interventional cardiology is a dynamically evolving field of medicine and represents the most frequent revascularization strategy used for the treatment of coronary artery disease in the United States. Over the past several years, numerous new medical devices have been introduced, which have greatly expanded the spectrum of patients that can be successfully treated. While the approval for the release of new devices is rigorously regulated, the post-marketing evaluation of safety is less clearly defined. Current practices for assessing the safety of medical devices depend on voluntary reporting by providers, and suffer from underreporting, temporal delays and the inability to assess rates of adverse events. We propose to develop a system to monitor the safety of new medical devices introduced into clinical interventional cardiology practice, utilizing a Bayesian updating strategy that addresses some of these deficiencies.
The Bayesian methodology incorporates existing knowledge regarding the safety of a medical device, typically from published clinical trial data, as well as observed data into the estimate of the risk of an adverse event. Using validated risk stratification methods, we propose to monitor the frequency of adverse events in order to assess the current state of risk of a procedure over time. We plan to evaluate the methodology on both historical and prospective data exploring the risks associated with the use of two emerging device classes within interventional cardiology: drug eluting stents, and filter-type distal protection devices. Preliminary exploration of the methodology for 309 patients who underwent rotational atherectomy demonstrates the feasibility of monitoring the evolution of risk estimates over the study period.
We propose to develop a formal theoretic framework for the establishment of prior probability estimates as well as methods for comparing results from the Bayesian system to that of classical statistics. We will then develop the safety monitoring analytic software, implementing both the Bayesian methodology as well as classical statistical methods, and deploy the system as a component of the outcomes tracking database system in existence in our facility. We propose a detailed analysis that will help determine if the Bayesian framework has any advantages over classical methods in monitoring device safety in interventional cardiology. Once validated, the analytic software will be made publicly available.