Geographically Equitable Liver Allocation Using Mathematical Redistricting

Abstract

Introduction: Geographic equity has proven elusive in the United States liver allocation system. Proposals for broader sharing between donor service areas (DSAs), which are grouped into regions, have not been approved. The regional groupings were not designed for equitable organ sharing. Methods: Mathematical optimization methods have been used to organize voting districts or school districts; here we apply them to design organ sharing regions within the United States. Inputs are donation and waiting list registration rates, as well as distances between DSAs. The resulting regional maps are tested using the Liver Simuated Allocation Model to find median MELD at transplant within each region and DSA, among other metrics. Results: Two alternative regional maps are proposed. Either of these alternative maps would reduce geographic disparities. The variance of median MELD at transplant among the DSAs is significantly lower using either of the alternative maps (by Levene's robust test statistic, p < 0.05). Note that although broader sharing within the existing regions would reduce predicted waitlist deaths from 1658.3 to 1600.8, it would actually worsen geographic disparity as demonstrated by a larger standard deviation of median MELD among DSAs.Figure: [Alternative Map 1]Figure: [Liver simulated allocation model results]Conclusion: Simply offering organs over a broader geographic area will not necessarily reduce geographic inequity in organ allocation. Summary metrics, like the number of people who die on the waiting list, could possibly improve at the same time that disparity metrics like the range of organ availability in different places worsens; allocation changes intended to address disparities must be evaluated using direct disparity metrics that measure variation. Mathematical redistricting may help reduce geographic inequity in allocation.