Inverse Decision Theory: Inferring Bounds on Costs Associated with a Current Standard of Care

• Activity Number: 416
• Type: Contributed
• Date/Time: Wednesday, August 6, 2003 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Stat. Sciences
• Author(s): Richard J. Swartz*+ and Dennis D. Cox and Michele Follen and Scott Cantor
• Companies: Rice University and Rice University and University of Texas and University of Texas
• Keywords: Bayesian sequential decision theory ; medical decision making ; revealed preferences

Abstract

This work develops a method to determine bounds on relative costs and losses for medical decisions. First the medical decision process is modeled as a Bayesian sequential decision problem. We model a finite sequence of diagnostic tests with a predetermined order. Then, to estimate bounds on the costs and losses, we assume the current standard of care for a particular disease is the optimal decision rule (i.e. it minimizes the Bayes Risk). Working backwards from the mathematical conditions resulting from this assumption, we determine the bounds on the costs and losses necessary to maintain this assumption. This method potentially incorporates both monetary considerations and patient outcomes into the costs and losses. We applied the method to data on the diagnosis of pre-cancerous lesions of the cervix from U. T. M. D. Anderson Cancer Center and compared the empirical bounds to values reported in the literature. This research was supported in part by The National Cancer Institute grant 2PO1-CA82710 and in part by a fellowship supported by the National Cancer Institute grant R25 CA57730, Robert M. Chamberlain, Ph.D., Principal Investigator.