• Exception;
  • liver transplantation;
  • MELD;
  • organ allocation;
  • UNOS region;
  • waiting list


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

MELD (model for end-stage liver disease) exception awards affect the liver allocation process. Award rates of specific nonhepatocellular carcinoma exceptions, termed symptom-based exceptions (SBE), differ across UNOS regions. We aimed to characterize the regional variability in SBE awards and examine predictive factors for receiving a SBE in the MELD era. The OPTN liver transplant and waiting list dataset was analyzed for waiting list registrants during the MELD allocation on February 27, 2002, until November 22, 2006. Competing risks proportional hazards regression analysis was used to examine predictors for receiving a SBE in 39 169 registrants. The hazard ratios for receiving a SBE differed significantly across regions when adjusted for multiple variables including age, gender, ethnicity, physiologic MELD score, blood group, functional status, etiology of liver disease, insurer and education level. Utilization of SBE is highly significantly variable across UNOS regions, and does not correlate with organ availability as estimated by the regional mean physiologic MELD score at transplantation. Patients with Medicaid as their primary payer have a lower likelihood of receiving a SBE award, while patients with cryptogenic/NASH cirrhosis or cholestatic liver disease have a higher likelihood of receiving a SBE. Reasons for these regional and demographic disparities deserve further investigation.


analysis of variance


hepatocellular carcinoma


hazard ratio


model for end-stage liver disease


MELD Exception Study Group and Conference


nonalcoholic steatohepatitis


Organ Procurement and Transplantation Network


regional review board


symptom-based exception


United Network for Organ Sharing


veterans affairs


waiting list death

Allocation of deceased donor liver grafts by MELD (model for end-stage liver disease) was instituted in the United States because of the lack of correlation between the length of waiting list time and waiting list mortality for liver transplant candidates (1–3). Several studies have demonstrated that both the number of new listed patients and waiting list mortality have decreased since the change to MELD allocation (4–6). MELD has also been studied as a predictor of pretransplantation mortality and short-term post-transplantation outcome (7–12).

A MELD exception award is an addition of points to a patient's laboratory-based MELD score or award of a standard MELD score in conditions where a patient's need for transplantation may be governed by other factors not accurately reflected by the physiologic MELD score (13). MELD exception awards have been most notably employed in patients with hepatocellular carcinoma (HCC) (14–19), but exception points may also be awarded for other conditions such as specific inherited metabolic disorders and hepatopulmonary syndrome. Additionally, MELD exceptions may be requested for symptom-based complications of severe chronic liver disease and portal hypertension that may affect the predicted waiting list survival of the candidate. These exceptions, which we have termed symptom-based exceptions (SBE), are submitted by a center's transplant team on behalf of an individual patient to their respective UNOS Regional Review Board (RRB). RRBs evaluate exception requests and subsequently award or deny additional points on a case-by-case basis by a majority vote of board members. RRBs are composed of physician and surgeon representatives of each region's transplant centers. Table 1 provides a list of indications for standard exceptions and symptom-based MELD exceptions.

Table 1.  Standard and symptom-based MELD exceptions
Standard exceptionsSymptom-based exceptions
Familial amyloid polyneuropathyRefractory ascites
Primary hyperoxaluriaRecurrent cholangitis
Hepatopulmonary syndromeRefractory encephalopathy
Small-for-size syndromePruritis Gastrointestinal hemorrhage

Prior to the MELD era, there was evidence of regional variability in the number of requests for and approvals of SBEs (20,21). The presence of regional variability suggests that the criteria for SBE request and approval differ across regions. This study has three main aims. First, we examine the current regional SBE request and award patterns in the MELD era. Second, we evaluate possible predictors for individual transplant candidates to receive a SBE. Third, we compare SBE likelihoods across regions to investigate for trends as they relate to markers of regional organ availability.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

The US Organ Procurement and Transplantation Network (OPTN) database was queried to establish the study cohort which comprised patients who were registered for liver transplantation between February 27, 2002, and November 22, 2006. Changes in definitions for MELD exceptions or the criteria used to make award decisions for MELD exceptions (MESSAGE committee guidelines [22]) made after 2006 may have altered differences in regional practices identified in this analysis using data from 2002 to 2006. Patients were excluded if they were less than 18 years at listing. Patients were also excluded if they underwent living-donor liver transplantation, transplantation after status 1 listing, retransplantation or multiorgan transplantation as these scenarios occur outside of MELD allocation principles. Patient-level data were used to calculate regional mean physiologic MELD scores at transplantation in patients transplanted without a MELD exception and in those transplanted with a SBE.

A SBE was defined as an award of additional MELD exception points for the following indications: ascites, bacterial cholangitis, hepatic encephalopathy, gastrointestinal bleeding or severe pruritus. These conditions were previously categorized under the “other specify” category of exception diagnoses prior to the recently published MESSAGE criteria.

The dataset contained patient-level information collected for all registrants for liver transplantation as well as concerning all exception requests and awards considered by the 11 UNOS liver RRB since the inception of MELD allocation. All MELD exceptions were initially stratified into categories of HCC and non-HCC followed by further stratification into categories of standard and SBE. Regardless of transplantation/waiting list or exception status, all patients meeting the inclusion criteria (including those with HCC exceptions) continued to be part of the overall study population in all analyses. The MELD score utilized in this analysis is the physiologic MELD score at registration. These analytical practices are consistent with other studies examining aspects of liver graft allocation and MELD (4,5,7,8,23).

Competing risk regression analysis was used to estimate the probability of each registrant's likelihood of being awarded an SBE while awaiting transplant (24). Competing risks analysis is an adaptation of Cox proportional hazards regression that recognizes that alternate events may be of similar or greater clinical significance than the event of interest. Cox proportional hazards regression, which produces hazard ratios to assess the effects of confounders and variables of interest on an event, is the traditional time-dependent analysis to quantify the hazard or likelihood of the occurrence of one event or for multiple independent events. In the scenario of SBE awards, transplantation and waiting list death, however, it is clear that certain events are not completely independent of one another, thus violating the assumption of independence of events required for classic Cox analysis. For example, undergoing transplantation or suffering waiting list death prior to SBE completely eliminates a candidate's ability or likelihood to receive a SBE; thus, they are not independent. Conversely, it is evident that receiving a SBE significantly impacts a candidate's hazard or likelihood of undergoing transplantation or receiving other MELD exceptions. Because of these dependent relationships between potential events, competing risks analysis represents a better choice for this multievent analysis. Censoring for the different events that may occur is done in an informative manner in competing risks analysis, rather than in a blind manner as in traditional Cox modeling and Kaplan–Meier survival estimation. Competing risks analysis utilizes the cumulative incidence as well as individual event incidences to provide both time-to-event estimates but also event-specific hazard ratios for the different events adjusted for their interdependence (25,26). Competing risks analysis has been demonstrated to provide improved estimates of the effects of patient characteristics on liver transplant waiting list mortality (25) and event risk estimates calculated for other study populations (26,27) compared to traditional Cox modeling and Kaplan–Meier estimation. In order to compare the effects of the two modeling techniques on our results, we performed the analysis separately using each technique independently and provide both results.

Cumulative incidence functions (probabilities of each event occurring over the time period of interest) were estimated for four competing events: receipt of an SBE award, death while awaiting transplant or removal from the wait list due to illness, receipt of a liver transplant and receipt of other exception awards. Hazard ratios for the occurrence of an SBE award were estimated accounting for these competing risks, with patients censored according to the first event that occurred during their course. Multivariable competing risk regression, based on the proportional subdistribution method, was conducted with adjustments for the primary covariate of interest, UNOS region and for the simultaneous effects of the following patient characteristics: MELD score, registration year, encephalopathy severity, ascites severity, functional status, liver disease diagnosis, blood group, age, ethnicity, insurer, education level, gender and foreign birth (24). The statistical significance of the adjusted effects of each covariate was assessed at the p < 0.05 level. SAS 9.2 and R were used for all data management and analysis calculations.

Of the 40 899 patients who met inclusion criteria, 39 169 had complete datasets for model variables and were included in the competing risks proportional hazards regression models (4.2% of patients were missing data). No data imputation was performed in the final model given the low prevalence of missing data.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

MELD-era candidates

Table 2 lists the descriptive characteristics of the study population. 39 169 listed candidates had complete datasets since the inception of MELD through November 22, 2006. Of the five possible first-events, 14 325 underwent transplantation, 7845 candidates died or removed from the list due to severity of illness and 10 561 candidates continued on the waiting and had no censoring event. There were a total of 6438 first-event MELD exceptions: 4705 were HCC-related, 391 were standard exceptions and 1342 were SBEs. Across all regions, 3.4% of the study population received SBE exception awards. The region-specific percent ranged from 1.5% to 6.2%.

Table 2.  Study population characteristics
 Frequency (or mean)Percent (or standard deviation)
  1. SBE = symptom-based exception; VA = veterans affairs; NASH = nonalcoholic steatohepatitis.

Total UNOS registrants for liver transplantation, 2002 through 200639 169100.00
First event during follow-up  
No event/on waiting list10 56126.96
SBE exception award 1 3423.43
Any other exception award 5 09613.01
Death/removal from list due to illness 7 84520.03
Transplant14 32536.57
Days from registration to first event386.9 days±492.4 days
UNOS region of listing for liver transplant
 Region 1 14593.72
 Region 2 531913.58
 Region 3 485812.40
 Region 4 423610.81
 Region 5 695517.76
 Region 6 12273.13
 Region 7 32208.22
 Region 8 22075.63
 Region 9 36199.24
 Region 10 28797.35
 Region 11 31908.14
Registration calendar year
 2002 712318.19
 2003 772519.72
 2004 822220.99
 2005 829021.16
 2006 780919.94
Encephalopathy severity
 Grade 3–4 4 84212.36
 Grade 1–220 96953.53
 None11 45329.24
 Missing 1 9054.86
Ascites severity
 Moderate to severe 9 60824.53
 Mild to moderate20 15751.46
 None 7 49819.14
 Missing 1 9064.87
 Other 1 9975.10
 African American 2 9687.58
 Hispanic 5 44413.90
 White28 76072.43
ABO blood group
 A14 54137.12
 B 4 76412.16
 AB 1 5253.89
 O18 10446.22
 Missing  2350.60
Insurer, primary payer
 Self-pay  3510.90
 Medicaid 5 90215.07
 Medicare 6 67717.05
 Government agency or VA 1 4453.69
 Private24 20961.81
 Missing  5851.49
Education level
 Missing/unknown10 93427.91
 Less than high school 1 6464.20
 Attended college or technical school 6 43716.43
 Associate or bachelors degree 5 97015.24
 High school14 18236.21
Diagnosis of liver disease
 Alcohol 6 11815.62
 Viral17 30644.18
 Cryptogenic/NASH 4 74512.11
 Cholestatic3 4038.69
 Drug  2090.53
 Neoplasm 1 5734.02
 Congenital  9382.39
 Vascular  1500.38
 Autoimmune  3480.89
 Missing 4 37911.18
Age52.0 years±10.0 years
Lab calculated MELD score19.6±10.4
 Male25 55865.25
 Female13 61134.75
Foreign born  
 Not foreign born37 96196.92
 Foreign born 1 2083.08
Functional status
 Unknown 5 99815.31
 Needs significant assistance 1 9625.01
 Needs some assistance10 05225.66
 Minimal or no assistance21 15754.01

Regional mean MELD scores and regional symptom-based exception approval rates

Table 3 displays the mean physiologic MELD score at the time of registration and at transplantation in each region over the study period. Regions 1 (22.1 ± 9.5), 5 (23.7 ± 11.2), 7 (21.3 ± 9.6), and 9 (20.9 ± 10.4) had the highest mean physiologic MELD scores at transplantation. The mean overall approval rate for all types of MELD exception was 74%. SBE approval rates ranged between 33% and 65% with a median acceptance rate of 53%.

Table 3.  Regional mean lab-MELD scores at listing and transplantation
UNOS region of listing for liver transplantMean lab-MELD at listingMean lab-MELD at transplant
  1. SBE = symptom-based exception; Lab-MELD = physiologic MELD score.

  2. p-Values for regional differences were calculated using single-factor ANOVA.

Region 118.3 ± 8.322.1 ± 9.5
Region 216.9 ± 7.419.7 ± 8.8
Region 317.5 ± 7.118.9 ± 7.8
Region 416.7 ± 7.519.3 ± 8.2
Region 518.4 ± 9.423.7 ± 11.2
Region 616.5 ± 6.818.7 ± 7.8
Region 718.0 ± 8.521.3 ± 9.6
Region 817.4 ± 7.720.7 ± 8.5
Region 917.7 ± 9.120.9 ± 10.4
Region 1015.4 ± 6.517.0 ± 7.4
Region 1117.4 ± 7.119.7 ± 7.9
All regions17.3 ± 7.820.1 ± 9.0

Outcomes of patients whose SBE requests were accepted

A total of 1342 candidates had a SBE as their first-event. Of those patients, 935 underwent transplantation (69%), while 407 did not. Of the 407 candidates who did not undergo transplantation, 228 (56%) suffered waiting list death or removal due to illness (17% of all candidates with first-event of SBE).

Outcomes of patients whose SBE requests were denied

Table 4 shows the rates for denial of SBE requests by region as well as the eventual outcomes of the patients who did not receive a SBE despite a request. A total of 48% of SBE requests were denied nationwide. Of the 31% of patients whose SBE request was denied eventually underwent transplantation per their physiologic MELD, 39% remained waitlisted without an exception and 30% died on or were removed from the waiting list. The mean physiologic MELD score in patients whose SBE request was denied was 20.1 at the time of database query.

Table 4.  Outcomes of patients in whom SBE request was denied
UNOS region of listing for liver transplantSBE denials (% of requests)Transplanted (%)Wait listed (%)WLD (%)
  1. SBE = symptom-based exception; WLD = waiting list death or removal from list due to illness.

Region 166184834
Region 244254628
Region 350364123
Region 450344224
Region 547294328
Region 635333333
Region 744364024
Region 853304129
Region 947273538
Region 1043472528
Region 1148263638
All regions48313930

Model for receiving a symptom-based exception (SBE)

Table 5 includes the adjusted hazard of SBE award by region, referent to region 11 calculated from the competing risks regression model. Region 11 was assigned as the reference region because it had the median regional mean physiologic MELD at transplant. Individual candidates in region 9 (HR = 1.672, p = 0.0001) had the highest likelihood for receiving a SBE, and candidates in region 6 (HR = 0.398, p = 0.0003) and region 7 (HR = 0.477, p < 0.0001) had the lowest hazard ratios. For example, patients in region 9 had a 67% greater likelihood of SBE award, compared to patients in region 11, after accounting for differences in MELD score, registration year, severity and all other measured characteristics. Also of interest, patients with Medicaid as their primary insurer had a significantly lower likelihood of receiving a SBE award (HR = 0.73, p = 0.029), while candidates with cholestatic liver disease (HR = 3.859, p < 0.0001) or cryptogenic/NASH cirrhosis (HR = 1.683, p = 0.0001) had a significantly higher likelihood of gaining a SBE award.

Table 5.  Competing risk regression results
 Hazard ratio95% confidence intervalp-Value
  1. SBE = symptom-based exception; VA = veterans affairs; NASH = nonalcoholic steatohepatitis.

UNOS region of listing for liver transplant
 Region 10.526(0.344–0.805)0.0031
 Region 20.504(0.376–0.676)<0.0001
 Region 30.556(0.421–0.734)<0.0001
 Region 40.888(0.688–1.147)0.3600
 Region 50.825(0.646–1.053)0.1200
 Region 60.398(0.241–0.657)0.0003
 Region 70.477(0.342–0.666)<0.0001
 Region 80.731(0.541–0.988)0.0410
 Region 91.672(1.292–2.162)0.0001
 Region 101.327(1.028–1.714)0.0300
 Region 11Reference  
Registration calendar year
Encephalopathy severity
 Grade 3–4Reference  
 Grade 1–20.822(0.651–1.038)0.1000
Ascites severity
 Moderate to severeReference  
 Mild to moderate0.403(0.342–0.476)<0.0001
 African American0.829(0.640–1.074)0.1600
ABO blood group
Insurer, primary payer
 Government agency or VA1.032(0.726–1.466)0.8600
Education level
 Less than high school0.81(0.555–1.183)0.2800
 Attended college or technical school1.042(0.860–1.262)0.6800
 Associate or bachelors degree1.129(0.934–1.364)0.2100
 High school0.882(0.747–1.042)0.1400
Diagnosis of liver disease
  Age 500.986(0.980–0.992)<0.0001
Physiologic MELD score0.945(0.935–0.954)<0.0001
  Lab calculated MELD score 200.965(0.956–0.973)<0.0001
Foreign born
 Not foreign bornReference  
 Foreign born1.334(0.941–1.89)0.1100
Functional status
 Needs significant assistance0.724(0.487–1.075)0.1100
 Needs some assistance0.937(0.750–1.172)0.5700
 Minimal or no assistance0.888(0.732–1.077)0.2300

Figure 1 shows plots of the cumulative incidence functions for receipt of an SBE award, as well as for the competing events of death while awaiting transplant or removal from the wait list due to illness, transplant and receipt of other exception awards over the complete follow-up period. The quartile distribution of the number of days from registration to the occurrence of any of these four competing events was as follows: 22 days at the 25th percentile, 153 days at the median and 604 days at the 75th percentile. Figure 2 includes plots of hazard ratios to receive a SBE award by region of listing. Statistically significant adjusted differences in the likelihood of SBE exception awards occurred among most regions. Interestingly, hazard for receiving a SBE award did not correlate to mean regional MELD at transplant, commonly held as a surrogate for organ availability (r =−0.12, p = 0.73). This indicates that there is no significant association between the likelihood to receive a SBE award and organ availability by this measure.


Figure 1. Cumulative incidence functions for receiving a SBE and competing risks. This plot shows the cumulative incidence function for receipt of an SBE award as well as functions for the competing events of death while awaiting transplant or removal from the wait list due to illness, transplant and receipt of other exception awards over the complete follow-up period. The mean number of days from registration to occurrence of any of these four competing events was 386.9 days (standard deviation ± 492.4 days).

Download figure to PowerPoint


Figure 2. Hazard ratios by region for receiving a symptom-based exception. Region 11 is the reference region (median regional physiologic MELD at transplant). Number of observations = 39 169. Variables included in the final model (also reflected in Table 5 and Supporting Table S1) include age, gender, ethnicity, region of listing, year of listing, presence of encephalopathy, presence of ascites, ABO blood group, insurance payer, education level, primary liver disease diagnosis, functional status, physiologic MELD score and foreign birthplace.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Concerns about consistency in MELD allocation in liver transplantation have been present since institution of the policy. Several adjustments have already been made in MELD exception awards for HCC. Rates of “special case” exception awards have been previously characterized and their use was variable across regions (8,20,23). Sequelae of these regional differences, however, have not previously been investigated. We examined a specific subset of MELD exceptions, termed SBE, which include only the MELD exceptions awarded to patients with symptomatic conditions. SBE awards represent organ allocation outside the paradigm of MELD because the vast majority of patients who receive such an award receive a transplant despite their physiologic, lab-based MELD score being lower than others who remain on the waiting list. We show that regions with similar mean physiologic MELD at transplant, a traditional surrogate for organ availability, have different utilization rates of SBE awards.

There were marked dissimilarities in likelihoods of SBE awards within strata of regional organ shortage as defined by regional mean physiologic MELD at transplant. The regional hazard ratio for receipt of a SBE did not correlate with mean regional physiologic MELD at transplant. Because we hypothesized that SBE utilization chiefly was a response to organ shortage, we expected that the likelihood of receiving a SBE would be relatively similar in regions with similar mean physiologic MELD at transplant; however, our analysis shows that there are clear regional differences in SBE award practices irrespective of organ availability and that this hypothesis is likely to be incorrect. For example, regions with the highest mean physiologic MELD at transplant (regions 1, 5, 7 and 9) had a relatively similarly high mean physiologic MELD score at transplantation. Despite this similarity, these four regions had highly variable likelihoods for a patient to receive a SBE. Regions 1 and 7 have very low HRs for SBE (0.526 and 0.477), while region 9 has the highest rate of SBE approval (1.672) and region 5's HR for SBE is between the two (0.825). The differences in likelihood of a SBE award between the regions with high degrees of organ shortage are notable, particularly when receiving a SBE is protective for an individual candidate and usually results in that patient undergoing transplantation. Perhaps these differences represent differing philosophies in how conservative RRBs are in deciding which patients have higher likelihood of waiting list death despite similar physiologic MELD scores. These differences may also result from opposing regional strategies in SBE utilization and differences in regional criteria used to request and approve SBE awards.

We conclude that this variability represents distinct regional differences in SBE award practices. This concedes the assumption of similarities of local factors such as regional exception criteria agreements and unique regional characteristics such as population density, waiting list epidemiology, liver donation and organ utilization, and listing practices. The interactions of these difficult-to-measure factors within the transplantation framework are unique to each region. Obviously this provides little granularity concerning the effects of these regional factors; conversely, it provides many interesting questions with regards to the impact of individual regional factors on the probability of events like transplantation and SBE within regions and within strata of organ availability. Because receipt or denial of an SBE request potentially has a large impact on a patient's likelihood of undergoing transplantation, differences in SBE awards by region also impact the likelihood of an individual patient to receive a transplant by their region of listing, which subsequently raises further questions regarding access to liver transplantation.

Our study has several limitations. We used a large database with patient-specific information collected for research, institution-to-institution quality assessment and funding purposes, and thus information bias may be present if completeness of data reporting varied across regions. Certain variables have subjective components (i.e. degree of encephalopathy, severity of ascites) that may affect their interpretability across regions and may thus have affected the results. A small proportion of patients had missing data and thus was not included in the analysis. We did not detect a systematic basis for data omissions from the analysis with performance of chi-square analysis of demographic variables of the excluded patients.

We specifically bounded our time period for this analysis to begin with the initiation of MELD allocation policy in 2002 and to end with the institution of the MESSAGE criteria in 2006 in order to avoid events with a possibility of causing sudden and meaningful changes in center and regional policy and behavior toward SBE awards. Unfortunately several policy changes related to HCC-related MELD exception awards also occurred during the analysis period, and patient registration and transplant decision-making at both the center and region level may have been affected by these changes, which occurred in 2003, 2004 and 2005. The effects, if any, of these policy changes were not adjusted for in our analysis.

Another potential limitation of this analysis is that regional criteria for competing events that preclude SBE award, including receipt of transplant or receipt of other exception awards, might vary meaningfully along with differences in SBE criteria. The potential variation in the likelihood of competing events associated with differences in regional criteria is not addressed in the analysis.

One other feature of note concerning our analysis is the use of competing risks analysis.

Prior to receiving an SBE award, patients can experience other key outcomes including death while awaiting transplant, removal from the waiting list due to illness, receipt of transplant or receipt of other exception awards. These alternative outcomes are competing risks for patients seeking SBE awards while on the waiting list, since the occurrence of one precludes the others. We selected competing risks analysis rather than more traditional Cox proportional hazards regression due to the interdependence of the possible events as well as the informed nature of censoring that competing risks utilizes.

While it is not completely clear, it does not appear that greater SBE utilization is a consistent cause of or response by RRBs to higher-than-average waiting list mortality, and both scenarios may be at play in the complex picture of liver transplantation. It is clear from our results that SBE awards are not employed uniformly across all regions with similar organ availability.

It is widely acknowledged that the proportion of transplants performed in patients with HCC exceptions, which still make up a minority of those who undergo transplantation, has far-reaching downstream effects on the threshold for transplantation for the group of waiting list patients as a whole (28–31). These effects have resulted in multiple adjustments to HCC exception awards since the inception of MELD. By this rationale, SBE awards also affect the likelihood of receiving a transplant for patients without a SBE award, albeit to a lesser degree than HCC exceptions. Some of the additional predictors that contribute to inter-regional variability likely include differences in patient demographics and dissimilarities in regional practice methods such as in criteria for organ acceptability for transplantation, listing practices and population density and distribution (urban, rural or suburban). We have attempted to minimize these factors by comparing regions with similar degrees of organ availability as estimated by mean physiologic MELD at transplant; however, several of these factors are difficult to measure and account for in a completely objective manner.

There are additional findings from our study that merit further investigation. In addition to the regional disparities in SBE approval, we also found that differences in aspects of socioeconomic status and liver disease etiology had significant impacts on the likelihood to receive a SBE in our adjusted model. The disproportionate rates of SBE awards by insurance payer, specifically in patients with Medicaid as their primary payer, and by etiology of chronic liver disease, namely cryptogenic cirrhosis and cholestatic liver disease, should prompt further evaluation of the effectiveness and fairness of the current system for granting MELD exceptions and the factors resulting in these demographic biases. Areas for future research include examining potential barriers to liver transplantation access as well as assessing behavioral differences in patients and providers in considering exception diagnoses or appropriate alternate indications for MELD exceptions.

These results suggest that further standardization of SBE approval criteria should be investigated starting first with an interim assessment of the impact of the MESSAGE consensus conference on waiting list mortality and SBE award rates. A national review board for SBE awards would effectively standardize exceptions across regions, but this step would reduce regions’ ability to adjust their exception criteria according to the types of candidates they care for. From our results concerning regional differences among strata of organ availability, there are undoubtedly differences in the types of patients, relative severities of illness and philosophies of members of the transplant community across regions. More investigation of additional factors accounting for the differences in waiting list mortality across regions may recommend other ways to maximize consistency in MELD allocation of deceased donor liver grafts without increasing regional disparities in access to transplantation.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

The authors thank Viktor E. Bovbjerg, PhD, for his thoughtful comments on the manuscript. This work was supported in part by Health Resources and Services Administration contract 234-2005-370011C. The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the US Government. Portions of this research were presented in poster format at the AASLD 2006 Annual Meeting held in Boston, MA.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Table S1: Cox proportional hazards regression results

AJT_3738_sm_TableS1.docx107KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.