Progress in healthcare research depends heavily on demonstrations of efficacy and safety via clinical trials. Bayesian methods offer a valuable alternative mode of analysis for clinical trials. By comparison with standard frequentist methods, Bayesian methods provide more interpretable outputs, use all the available evidence, and lend themselves to the more complex analyses demanded by complex healthcare research. Unfortunately, few biostatisticians have received training in Bayesian methods. We propose to produce user-friendly software to enable Bayesian analysis of data from clinical trials. This software will implement the complete functionality of Spiegelhalter s BART software, but in a much easier-to-use package. We will develop an associated web-based tutorial. In Phase II we will develop the software into a full-featured Bayesian analysis toolkit for clinical trialists.
Clinical trials represent a 10 billion dollar industry. Statistical software for this industry is itself a significant business. We sell iBART both to drug company trialists and to federally sponsored clinical researchers. We expect iBART s tutorial and the Bayesian approach in general to appeal especially to the many trialists without specialized statistical qualifications.
