PURPOSE: Currently, 59 Organ Procurement Organizations (OPOs) are aggregated to 11 regions in the U.S. organ transplantation and allocation network: allocation rules include a regional preference. Given this hierarchy, we seek to assess whether the current regional configuration optimizes transplant efficiency, i.e., meets a set of criteria such as maximizing transplant cardinality and minimizing organ wastage. A prior analytic framework proposed by Stahl revealed that the current configuration is sub-optimal, however, it was computationally intractable to fully optimize over all possible regional configurations. We have adapted a solution method that allows models of sufficient size and complexity to be solved, and have added additional parameters to enhance clinical realism.

METHOD: Integer programming (IP) is an industrial engineering technique that deals with problems making discrete choices from a potentially large set of possibilities: in this case choosing the optimal regional configuration from a nearly infinite set of possibilities. Our analysis was conducted from the societal perspective and the allocation system was assumed to be steady. We modified an analytic proxy of the total number of intra-regional transplants to capture the effect of organ wastage on the region design. We adapted Branch and Price (B&P), an IP solution technique, to solve models that were previously intractable. B&P allowed us to examine region designs that prior techniques did not allow (regions composed of more than 7 OPOs), and to search the optimal configuration more efficiently. Data were obtained from UNOS or provided in the literature. We performed a univariate sensitivity analysis where we varied the percentage of organ wastage.

RESULTS: Using liver allocation as an example, if no livers wasted during allocation, a configuration can be found that provides an additional 13 livers per year. The average CIT would be decreased by 0.46 hrs. If there were liver wastage (the percentage of liver wastage varied from 5% to 40%), more additional livers would be transplanted (150 ~ 400 additional transplants). Compared to the current configuration, the optimal set of regions contains fewer regions of large size.

CONCLUSIONS: Our analysis indicates that efficient organ allocation may benefit through region reorganization. More importantly, the application of this analytic framework allows for models of realistic modeling complexity to be evaluated and optimized.