WHEN TO ACCEPT A CADAVERIC LIVER OFFER FOR TRANSPLANTATION? A MARKOV DECISION (MDP) MODEL APPROACH

To Register SMDM Homepage

Tuesday, 19 October 2004 - 8:30 AM

This presentation is part of: Oral Concurrent Session B - Methods in Judgment and Decision Making

WHEN TO ACCEPT A CADAVERIC LIVER OFFER FOR TRANSPLANTATION? A MARKOV DECISION (MDP) MODEL APPROACH

Oguzhan Alagoz, PhD¹, Andrew Schaefer, PhD¹, Lisa Maillart, PhD², and Mark S. Roberts, MD, MPP³. (1) University of Pittsburgh, Department of Industrial Engineering, Pittsburgh, PA, (2) Case Western Reserve University, Weatherhead School of Management, Cleveland, OH, (3) University of Pittsburgh, Section of Decision Sciences and Clinical Systems Managment, Pittsburgh, PA

Purpose: Although there is a shortage of cadaveric livers, 45% of all liver offers are initially declined by the transplant surgeons/patients. Organs are rejected in the hope that they will receive a better organ offer in the future. Our objective is to solve the decision problem faced by these patients: should an offered organ of a given quality be accepted or rejected? The decision depends on two major components: the patient's current and likely future health as well as the current and future prospects for organs. This extends our previous work that determined when to accept a living-donor liver. Methods: Markov decision processes (MDPs) are analytic tools for sequential decision-making under uncertainty. We developed an infinite horizon discounted MDP model that determines whether to accept or decline a liver offer. The state of the system is described by patient health and organ quality. The Model for End-Stage Liver Disease (MELD) score, a risk prediction score, represents patient health. We estimate the transition probabilities using the natural history of liver disease and the national liver offer rates. We found the optimal policies using a policy iteration algorithm. Results. The figure shows an optimal policy based on a simplified definition of organ quality, where the quality is determined by donor characteristics such as age, gender, race...etc. Organs are ordered in a decreasing quality, that is, Organ 1 is the best and Organ 14 is the worst organ. Note also that the higher the MELD score, the sicker the patient is. As expected, the optimal accept-decline decision is a function of organ quality and patient health. For example, when the patient has a MELD score of 18, the policy suggests that the patient should not accept liver offers that have a lower quality than 5 (that have a higher figure than 5). Conclusions: Given functions that represent patient illness and organ quality, this analytic model can be used to determine the accept/decline decisions for cadaveric organs.

See more of Oral Concurrent Session B - Methods in Judgment and Decision Making
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)