Exploration of a Bayesian Updating Methodology to Monitor the Safety of Interventional Cardiovascular Procedures

Abstract

Appropriate methods for monitoring of the safety of medical devices introduced into clinical practice have been elusive to develop and implement. A novel approach is the application of Bayesian updating, which incorporates existing knowledge regarding event rates into the estimation of risk. This framework has been shown in other domains to be data efficient and to address some of the limitations of conventional statistical methods. In this article, the authors propose a methodologic framework for developing initial prior probability distributions in risk-stratified patient groups and a mechanism for incorporating accumulating procedure safety experience. In addition, they use this methodology to retrospectively analyze the clinical outcomes of 309 patients undergoing an infrequent interventional cardiology procedure, rotational atherectomy. These exploratory analyses demonstrate the feasibility of Bayesian updating applied to medical device safety evaluation and indicate that the methodology is capable of generating stable estimates of risk in a variety of patient risk groups.