In the United States, the determination of which patient will be offered a liver for transplantation is based on 2 variables: allocation and distribution. Allocation is defined as the ordering of the patients on a particular waiting list. Distribution can be defined as the ordering of the waiting lists to which the organ is offered. For example, the Model for End-Stage Liver Disease (MELD) score and the waiting time are currently used for ranking patients for allocation on a particular waiting list.1
The current distribution system is based on the local unit (the donation service area) and regional and national tiers (Fig. 1). These geographic boundaries are to some degree arbitrary, are not uniform across the country, and are not based on any metric related to liver distribution. The Department of Health and Human Services issued what has been termed the final rule, which is the basis for transplantation policy in the United States. The policy states that 1 allocation performance goal is “distributing organs over as broad a geographic area as feasible … and in order of decreasing medical urgency.”2 The criteria for medical urgency have been successfully defined by the MELD score. The rationale for a distribution performance goal could be primarily the minimization of wait-list deaths and secondarily the reduction of geographic disparities in MELD scores at transplantation. The final rule explicitly states that organs should be allocated to the patients with the greatest medical need, which is currently defined as wait-list mortality and is measured with the MELD score for liver patients.
Disparities in MELD scores at transplantation and in death rates across the United States (Figs. 2 and 3) led to a proposal to extend the broader, regional distribution of cadaveric liver allografts to all patients listed with a MELD score; this proposal was distributed for public comment in 2009. Support for this policy was mixed and did not meet the threshold of acceptance in the community. Sponsored by the United Network for Organ Sharing Liver and Intestine Committee, a public forum was subsequently held on allocation and distribution in April 2010. Leading up to this forum, a great deal of allocation modeling was performed to evaluate the impact of potential changes to distribution if a policy were adopted. Liver simulated allocation modeling (LSAM)4 allows for possible changes to be made in allocation/distribution rules with anticipated results. LSAM is a simulation program that estimates the clinical impact of an alternative allocation system on key transplant outcomes, including the performance of transplantation, wait-list death, posttransplant death, and overall survival over a calendar year. LSAM evaluates data from actual candidate registrations and available donors and simulates the results according to programmed allocation systems.
Figure 4 shows the relationship between the median distance that an organ travels and the decrease in total deaths in a simulated year. Total deaths include the deaths of patients on the wait-list and the deaths of patients in the posttransplant period. In the current system, the median travel distance of a liver is 65 miles. If the distribution system were changed to a regional share MELD system (ie, the initial distribution unit would be the region rather than the local area, and the allocation system would use the MELD score), the median travel distance of an organ would increase to approximately 127 miles, and the number of total deaths would decrease by 61. In general, the greater the median travel distance is, the greater the decrease in total deaths.
A concern about increasing the distance that an organ travels is the effect of travel on the cold ischemia time. Work by the Scientific Registry of Transplant Recipients (SRTR) has found that there is a complex interaction between the cold ischemia time and the distance that an organ travels (Fig. 5). A partial explanation may be that there is a cold ischemia time considered to be acceptable for a specific liver allograft, and transplants may be delayed for logistical reasons if the cold ischemia time is within this parameter; for example, doctors may wait to perform transplantation early in the morning rather than in the middle of the night. If an organ is being shipped and a delay will make the cold ischemia time longer than desired, immediate transplantation will result without waiting until morning. These behavioral issues may vary with longer distances, so cold times are difficult to anticipate and model.
The concept of Net Benefit allocation has been published and presented previously.5 In essence, posttransplant outcomes are taken into consideration along with pretransplant mortality. There was a significant amount of interest expressed in this concept. The development of Net Benefit has not been completed, and concerns about the predictability of posttransplant survival and the complexity of the number of involved variables have hindered this from moving forward in policy development.
Prior work has shown the negative survival benefit that many patients experience with transplantation if their MELD score is <15.6 Subsequent policy adjustments have redirected local cadaveric livers regionally if there is no local patient with a MELD score ≥ 15 and then back to the local unit if there are no regional patients before national placement. One small incremental adjustment would be to extend this from a local-regional-local algorithm to a local-regional-national algorithm (national share 15). Modeling suggests that a national share 15 policy would result in a reduction of 50 total deaths annually. If we draw upon the generally favorable experience with regional sharing for status 1 patients, the extension of this type of sharing arrangement to sicker, non–status 1, high MELD score patients could be a reasonable and logical next step in a process to decrease wait-list mortality. If these 2 changes (national share 15 and regional share 35) were combined, modeling data suggest a reduction in total deaths of approximately 60, and the median distance between the donor hospital and the transplant center would increase from 65 to 85 miles; this suggests little effect on cold ischemia times or lengths of stay. The overall sharing of livers outside the local donation service area would increase from 29% to 36% (Fig. 6).
Just as the performance of transplant centers has been under scrutiny with respect to patient outcomes, the performance of organ procurement organizations (OPOs) is being evaluated with respect to organ donor yields. For each type of organ (heart, lungs, liver, kidneys, and pancreas), the observed yield is defined as the number of organs eventually transplanted into recipients. The observed yield is evaluated with respect to the expected yield, which is calculated with a statistical model developed by the SRTR. The initial criteria used to identify OPOs with lower than expected organ yields for all organs as well as each organ type will include all of the following:
More than 10 fewer observed organs per 100 donors versus expected yields [(observed organs per 100 donors) − (expected organs per 100 donors) < −10].
An observed yield/expected yield ratio < 0.90.
A 2-sided P value < 0.05.
Each OPO will be evaluated every 6 months for each type of organ with respect to the yield. Early estimates show that 5 to 8 OPOs will be flagged for review overall, or 1 to 2 OPOs will be flagged for each type of organ (kidneys, liver, pancreas, heart, and lungs).
There is a developing disconnect between OPOs and transplant centers. Because OPOs are being driven to maximize the potential of standard and nonstandard donors, transplant centers are being held strictly accountable to patient outcomes. Centers have little incentive to be aggressive with marginal donors, whereas OPOs need and want to get more organs transplanted to improve their yields. The measurements of expected donor yields will likely push OPOs to pursue all potential donors, regardless of whether local centers are interested. This could be advantageous to expanding the distribution of donor livers. Recent data suggest that livers placed nationally perform well when they are risk-adjusted.7 Improving yields and broadening distributions may help to mitigate differences in MELD scores at transplantation and in deaths on the wait list.