- Email: [email protected]
Liver Transplant Waiting Time Does Not Correlate With Waiting List Mortality: Implications for Liver Allocation Policy
RAPID COMMUNICATION
Liver Transplant Waiting Time Does Not Correlate With Waiting List Mortality: Implications for Liver Allocation Policy Richard B. Freeman, Jr,* and Erick B. Edwards,† for the United Network for Organ Sharing Liver and Intestine Committee Factors associated with the risk for mortality once placed on the liver transplant waiting list and how this risk relates to center-specific waiting time and transplant activity have not been adequately evaluated. We performed this study to determine the association between center-specific waiting time and waiting list mortality among liver transplant candidates stratified by medical urgency at the time of registration. A Cox proportional hazards model was used to calculate 2-year mortality risk for a cohort of 16,414 registrants added to the United Network for Organ Sharing liver transplant waiting list between January 1, 1997, and December 31, 1997. After controlling for confounding variables, we calculated the mortality risk for centers, organ procurement organizations (OPOs), and states. The relation between center-specific waiting list mortality risk and median waiting time or transplant activity was determined by linear regression. In multivariate analyses, higher initial medical urgency status (relative risk [RR] ⴝ 12.8; P < .001), increasing age (P < .001), black ethnicity (RR ⴝ 1.29; P < .001), history of previous transplant (RR ⴝ 1.2; P ⴝ .009), certain liver diagnoses, and smaller center size (RR ⴝ 1.39; P ⴝ .008) were associated with significantly increased waiting list mortality. Candidates with blood type A (RR ⴝ 0.87; P < .001) and those with cholestatic cirrhosis as the primary diagnosis (RR ⴝ 0.73; P < 0.001) had a reduced risk for dying. There were significant variations in 2-year waiting list mortality risk among centers, OPOs, and states. However, when stratified by medical urgency status at waiting list entry, center-specific waiting time and transplantation rates accounted for almost none of the center-specific waiting list mortality. Although there are variations in waiting list mortality risk among centers, OPOs, and states, there is very little relation between center-specific waiting list mortality and center-specific median waiting time or center-specific transplantation rates when stratified by medical urgency. Waiting time and center transplant rates should not influence liver allocation policy. (Liver Transpl 2000;6:543-552.)
O
ver the last few years, donor organ allocation policy, and liver allocation in particular, has been intensely scrutinized and criticized in the press and by the federal government.1 Much of the criticism has focused on the wide variations in the time that candidates for liver transplantation must wait for a donor organ and the geographic basis of these differences.2 In answer to these concerns, many liver transplant profes-
sionals have argued that waiting time for a liver transplant is not a good measure of the system’s performance. In 1998, in response to concerns that new proposed federal regulations for national organ allocation policy1 were unduly emphasizing waiting time as a liver allocation measure, Congress commissioned the Institute of Medicine (IOM) to study the liver allocation system.3 In addition, a recent preliminary government study reported wide variations in center-specific liver transplantation rates and implied that this variation might be caused by deficiencies in the allocation system.4 No other analyses of liver transplant waiting list mortality are available in the literature. In designing the present study of liver transplant waiting list outcomes, we chose a relatively uniform cohort of patients to be studied, obtained from a period during which liver allocation policy changed very little. The need to maintain this relative uniformity had to be balanced with the necessity of having an adequate follow-up period to reasonably determine the fate of patients entered onto the waiting list. This study focuses on waiting list mortality as a primary outcome because a more relevant concern is not how long a patient will wait, but what the risk is that a patient will die before getting an opportunity for transplantation. In addition, because patients with liver failure are referred to and managed by transplant centers, not organ procurement organizations (OPOs) or states, we chose to emphasize From the *Department of Surgery, Division of Transplantation, Tufts University School of Medicine/New England Medical Center, Boston, MA; and †United Network for Organ Sharing, Richmond, VA. Presented in part at the 50th Annual Meeting of the American Association for the Study of Liver Diseases, Dallas, TX, November 7, 1999. Address reprint requests to Richard B. Freeman, Jr, MD, Division of Transplantation, New England Medical Center, Box 40, 750 Washington St, Boston, MA 02111. Telephone: 617-636-5592; FAX: 617-6368228; E-mail: [email protected] Copyright © 2000 by the American Association for the Study of Liver Diseases 1527-6465/00/0605-0003$3.00/0 doi:10.1053/jlts.2000.9744
Liver Transplantation, Vol 6, No 5 (September), 2000: pp 543-552
543
544
Freeman and Edwards
center-specific analyses of waiting time and mortality risk. We also chose to define waiting time as the total time a patient spent on the list and stratify our analysis by initial medical urgency status at the time of entry onto the list. This is in contrast to the IOM method in which waiting time was defined as the amount of time a patient spent at a given medical urgency status while that patient was an active candidate.5 This aim of this report is to analyze the risk factors associated with liver transplant waiting list mortality and how center-specific waiting list mortality is related to center-specific waiting time and transplantation activity.
Methods Our study cohort consisted of 16,414 liver transplant candidates with registrations received by the United Network for Organ Sharing (UNOS) between January 1, 1996, and December 31, 1997. We intentionally chose this cohort of patients from a recent 2-year period during which liver allocation policy was stable and relatively current. In addition, this cohort had sufficient follow-up time to accurately determine the risk for death within 2 years of registration. Our primary outcome variable was death within 2 years of registration. Patients who were removed from the waiting list, underwent transplantation, or still waiting at the time of the analysis were censored. Patients removed from the waiting list for being too sick to undergo transplantation were scored as deaths on the list. These patients’ dates of death were randomly assigned to be 0 to 30 days from their date of removal from the waiting list. Seven distinct covariables were examined in a multivariable analysis to determine the relative risk (RR) for waiting list mortality. These were: (1) UNOS medical urgency status at the time of registration, (2) candidates’ blood group, (3) history of previous liver transplant, (4) candidates’ race, (5) age at registration, (6) primary diagnosis at listing, and (7) size of the center listing patients. Center size was defined as the number of new additions to the waiting list entered by each center during the study period. For center-size comparisons, centers were divided into quartiles based on their total new additions to the list. A Cox proportional hazards model was constructed to estimate the RR for waiting list mortality for the various patient and institutional factors previously identified. The proportional hazards assumption was assessed graphically for the statistically significant patient factors. Factors not satisfying this assumption were either dropped from the analysis or included as stratification factors in the Cox model. Once the final set of patient factors was identified, tests of significance were constructed for the institutional factors while simultaneously adjusting for significant patient factors. We did not include patients entered at the lowest medical urgency status (status 4) in our center-specific analyses. This category was eliminated in 1997 because essentially no patients in this
status level underwent transplantation and therefore is irrelevant to current practice. To assess the relation between waiting time and RR for waiting list mortality, we stratified our analyses by initial medical urgency status at the time of entry onto the list. This was done to address the criticisms of previous studies in which no stratification was performed and to comply with the broad opinion of transplant professionals that medical urgency must be taken into account when comparing waiting times to liver transplantation. Median waiting time for patients stratified by initial medical urgency status was determined by calculating the time at which 50% of all patients initially listed at a given medical urgency status received a transplant at each center. In this study, in contrast to the analysis of the IOM, waiting time was patient specific and cumulative. For example, waiting time for patients initially listed at medical urgency status 3 at a given center was calculated from each patient’s initial list date to their corresponding transplantation date regardless of other changes in medical urgency status between the list date and transplantation date. Similarly, for patients initially listed at status 1 or 2, waiting time was calculated from the initial list date to transplantation date, regardless of whether the patient’s medical urgency status changed between the list date and transplantation date. Conversely, the IOM analysis used the time all patients spent at a given medical urgency status to calculate status-specific waiting times regardless of whether a given patient was initially listed at that status. In the IOM calculations, patients who accrued waiting time at a given medical urgency status and then moved to another medical urgency status were censored in the analysis of waiting time for the first status.5 Each center’s patients were then stratified by medical urgency status at the time of listing, and their RR for death within 2 years of registration on the waiting list was calculated. Each center’s median waiting time was plotted against the RR for waiting list mortality for each center’s patients at each medical urgency status to determine the center-specific relation between waiting time and waiting list mortality. We also analyzed the relationship between the percentage of patients who underwent transplantation at each center and center-specific waiting list mortality risk within the 2-year study period. Again stratifying for medical urgency at time of entry onto the list, we plotted the RR for mortality against the percentage of patients listed who received transplants at the respective centers. This was done to assess the relationship between transplant activity and waiting list mortality for each center within each medical urgency category. A simple linear regression was used to determine these relationships. The results are plotted as scatterplots to better show the variation and provide a more complete picture of the data.
Results During our study period, 16,414 candidates were added to the UNOS liver transplant list. Study closeout date was May 9, 1999. All patients were followed up
545
Liver Transplant Waiting Time and Waiting List Mortality
until transplantation, death, or removal from the list, with 3,835 candidates (23.4%) still active on the waiting list at study closeout. One hundred fifteen centers from 52 OPOs added patients to the liver transplant waiting list. Candidates were from all 50 states. During our study period, prevailing organ allocation policy did not include minimum listing criteria,6 and entry onto the list at the lowest medical urgency was at the discretion of the transplant centers. Patients meeting the highest medical urgency status (status 1) were defined as those with acute or chronic liver disease who were confined to an intensive care unit on life support, with less than 1 week to live. Patients meeting the second most urgent status (status 2) were defined as requiring constant hospitalization. Those listed as status 3 required constant medical attention but not necessarily hospitalization, and status 4 included all others. The RRs for mortality once placed on the waiting list stratified by various patient factors are listed in Table 1. Patients entering onto the waiting list at the highest medical urgency (status 1) had more than a 12 times greater risk for death while on the list compared with those entering at the lowest 2 medical urgency
Table 1. Relative Risk for Waiting List Mortality Within 2 Years of Listing According to Patient Characteristics Factor Status at entry 1 2 3 Blood group A B, AB, O Previous OLT Race Black Other White, Asian, Hispanic Age at listing (yr) ⱖ60 v ⱕ45 Diagnosis at listing Cholestatic Laennec’s Noncholestatic Acute Malignancy Unknown Posthepatitic cirrhosis
Relative Risk
P
12.80 5.28 1.00
⬍.001 ⬍.001 Reference
0.87 1.00 1.20
⬍.001 Reference .009
1.29 1.59 1.00
⬍.001 .009 Reference
1.20
⬍.001
0.73 1.17 1.19 1.73 1.91 0.67 1.00
⬍.001 .003 .004 ⬍.001 ⬍.001 .012 Reference
Abbreviation: OLT, orthotopic liver transplantation.
Table 2. Relative Risk for Waiting List Mortality Within 2 Years of Listing According to Center Size Factor First quartile (0 ⬍ additions ⱕ 33) Second quartile (33 ⬍ additions ⱕ 106) Fourth quartile (181 ⬍ additions)
Relative Risk
P
1.39
.008
1.18
.002
1.00
Reference
NOTE. Center size is determined by number of new additions to the waiting list.
categories. Patients entering at medical urgency status 2 had more than a 5 times greater risk for death while on the list compared with those in the lowest 2 medical urgency categories. Patients with blood type A had a reduced risk for death while waiting compared with patients of other blood types. Patients with blood types B, AB, or O did not have significantly different mortality risks when individually compared with all other blood types combined (data not shown). Those with a previous liver transplant, of African descent, or of increasing age had slight but statistically significant increased risks for death on the waiting list. Patients with diagnoses of alcoholic, noncholestatic, or acute liver failure or hepatic malignancy had increased risks for death compared with patients with posthepatitic cirrhosis, whereas patients with cholestatic liver disease had a reduced RR for mortality on the list. New additions to the liver transplant list were tabulated for each liver transplant center during the study period, and the centers were then stratified in quartiles to determine the relation between the size of a liver transplant center’s waiting list and waiting list mortality (Table 2). Centers with the fewest additions to the list had significantly increased RRs for waiting list death compared with centers with the largest waiting lists. The RR for waiting list mortality for each center was adjusted for significant patient factors (waiting list entry status, race, age, blood type, and liver failure diagnosis) and plotted in Figure 1 to show the variation among the centers. Of the 115 centers, 14 centers had statistically significant increased RRs for death (RR range, 1.25 to 4.8) compared with all other centers, and 9 centers had significantly decreased RRs (RR range, 0.31 to 0.73). Figure 2 represents a similar analysis, plotting RR for waiting list mortality against candidates’ state of residence. Patients from 7 states had a statistically increased RR for death on the waiting list (RR range, 1.23 to
546
Freeman and Edwards
Figure 1. Relative risk for pretransplantation candidate mortality within 2 years of listing (vertical axis) shown by center (horizontal axis) for new additions to the liver transplant waiting list, 1996 to 1997. ( ) Centers with relative risks significantly different from all other centers. ID, identifier.
2.37) compared with all other states, and 3 states’ residents had a reduced RR for death (RR range, 0.77 to 0.83). For OPOs, patients listed with centers in 10 OPOs had significantly increased risks for death on the waiting list (RR range, 1.25 to 2.26), and 4 OPOs had a reduced risk for death on the list (RR range, 0.67 to 0.79; Fig. 3). Of the 14 centers with significantly greater mortality rates, 9 centers were in states that had significantly greater mortality rates. Eleven of the 14 centers were associated with OPOs that had significantly greater mortality rates. Of the 9 centers with significantly lower mortality rates, 6 centers were in states with significantly lower mortality rates. Six of the 9 centers were associated with OPOs that had significantly lower mortality rates. In Figure 4, each center’s median waiting time for
Figure 2. Relative risk for pretransplantation candidate mortality within 2 years of listing (vertical axis) shown by candidates’ state of residence (horizontal axis) for new additions to the liver transplant waiting list, 1996 to 1997. ( ) Centers with relative risks significantly different from all other centers. ID, identifier.
patients entering onto the list as status 1 during our study period are plotted against the RR for death on the waiting list for those same center-specific patients. There was no relation between the center-specific median waiting time and the center-specific RR for death for patients initially listed at status 1 (R ⫽ 0.03; R2 ⫽ 0.0009; P ⫽ .84). The few outlier points with extremely long waiting times were caused by a few centers listing only 1 or 2 patients as status 1 during our study period. These patients were downgraded to lower medical urgency levels, making their waiting times extremely long. When these outliers are excluded from the analysis, there is no significant change in the results (data not shown). To further address this phenomenon, we plotted the RR for death on the list for each center’s patients entering onto the list at status 1 against the percentage of those patients who underwent transplantation within
Liver Transplant Waiting Time and Waiting List Mortality
547
Figure 3. Relative risk for pretransplantation candidate mortality within 2 years of listing (vertical axis) shown by OPO (horizontal axis) for new additions to the liver transplant waiting list, 1996 to 1997. ( ) Centers with relative risks significantly different from all other centers. ID, identifier.
our study period at the respective centers (Fig. 5). Again, there is no relation between the center-specific RR for mortality on the list and the center-specific proportion of patients who underwent transplantation who were initially listed at status 1 (R ⫽ 0.033; R2 ⫽ 0.0011; P ⫽ .74). For liver transplant candidates initially entering onto the waiting list at medical urgency status 2, we found very similar results. There is no relation between the center-specific median waiting time and the centerspecific RR for death for patients initially listed at status 2 (Fig. 6; R ⫽ 0.11; R2 ⫽ 0.0117; P ⫽ .47). Again, several outlier points occurred because of the phenomena of a few centers initially listing a very small number of patients at status 2. Elimination of these outliers does not change the results of this correlation (data not shown). There was a very slight trend toward a reduced risk for death for patients listed at centers performing trans-
plantation on a greater proportion of their patients listed at status 2. Although this trend reached statistical significance, less than 5% of the variations in centerspecific mortality risk for status 2 patients can be attributed to the center-specific transplantation rate (Fig. 7; R ⫽ – 0.22; R2 ⫽ 0.0485; P ⫽ .024). For patients initially entering onto the list at status 3 (Fig. 8), there was a trend toward a reduced risk for center-specific waiting list mortality for patients listed at centers with the longer median waiting times (R ⫽ – 0.31; R2 ⫽ 0.0965; P ⫽ .004). This trend indicates a reduced risk for death for the longest waiting status 3 patients when analyzed on a centerspecific basis. For patients initially entering onto the list at status 3, there was no relation between the center-specific RR for mortality on the list and center-specific proportion of patients who underwent transplantation (Fig. 9; R ⫽ 0.08; R2 ⫽ 0.0074; P ⫽ .56).
Figure 4. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 1 (vertical axis) plotted against center-specific median waiting times (horizontal axis).
548
Freeman and Edwards
Figure 5. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 1 (vertical axis) plotted against center-specific percentage of patients listed who received transplants during the study period (horizontal axis).
Discussion Our analysis distinctly indicates several patient factors associated with increased risk for death on the waiting list. By far, the medical condition of the patients at the time of entry onto the list, defined by UNOS medical urgency status, was the most important factor overall associated with increased waiting list mortality. In our analysis, stratification of waiting list candidates by UNOS medical urgency criteria prevailing at the time of our study may not have been the most accurate measure of severity of liver disease. Nonetheless, that both our study and that of the IOM found waiting list mortality to be highly correlated with the UNOS medical urgency criteria indicates that these criteria have validity in predicting waiting list mortality (and therefore urgency for transplantation) for this cohort. The other patient variables associated with increased
Figure 6. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 2 (vertical axis) plotted against center-specific median waiting times (horizontal axis).
mortality risk were several times smaller in magnitude. Patients with a primary liver diagnosis of acute liver failure most often were in the highest medical urgency status, so it is not surprising that these patients also had an increased mortality risk. These data might be used to justify further direction of donor livers to the most medically urgent patients to try to alleviate some of their increased risk for death. However, on further examination, many other factors may be responsible for these findings. Having an available organ at the moment when these patients present to the waiting list is a very random event. Our data showing no more than a trivial relationship between median waiting time for status 1 (or status 2) patients and a negative relationship for status 3 patients and waiting list mortality support this notion. One might conclude that centers with longer median waiting times for status 1 patients might have higher (or lower) waiting list mortality risk if pa-
Liver Transplant Waiting Time and Waiting List Mortality
549
Figure 7. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 2 (vertical axis) plotted against center-specific percentage of patients listed who received transplants during the study period (horizontal axis).
tient disease progression and organ availability were both constant linear phenomena. Conversely, we found virtually no relation between waiting time and mortality for each medical urgency status, indicating that time on the list is not associated with an increased or decreased risk for death without a transplant. In addition, patients listed at centers performing transplantation on a high percentage of candidates initially listed at status 1 or 3 did not have a significantly increased (or decreased) risk for mortality on the list, and the variation in waiting list mortality risk attributable to center transplantation rates for status 2 patients was less than 5%. Therefore, our data do not directly support even more priority being given to these more critically ill patients to reduce their waiting time than was prevailing during our study period. That the other patient factors associated with increased risk for death on the list are relatively small
makes it difficult to advocate changes in the allocation system to equalize these risks. The effect of blood type on pre–liver transplantation mortality has been examined, with a slight but significant increase in pretransplantation mortality for candidates with blood type O compared with all other blood types.7 This report also found patients with blood type O waited longer for liver transplants and concluded that waiting time was associated with mortality. However, the data for this previous study came from only 3 large liver transplant centers and was culled from 1990 to 1993, when waiting times were dramatically shorter (median waiting time, 109 days for blood type O, 58 days for blood types A, B, and AB). Also, this analysis was not performed on a center-specific basis. The use of blood type as a surrogate for randomization to transplantation after longer or shorter waiting times also has limitations, as noted by the investigators of this study.
Figure 8. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 3 (vertical axis) plotted against center-specific median waiting times (horizontal axis).
550
Freeman and Edwards
Figure 9. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 3 (vertical axis) plotted against center-specific percentage of patients listed who received transplants during the study period (horizontal axis).
Factors such as age, race, and primary liver diagnosis are as much a function of referral patterns, insurance contracts, and demographics of referral areas as they are likely to be influenced by the allocation system. In the 1 report examining access to liver transplantation before entry onto the allocation list, demographic factors of age, sex, source of medical insurance payment, liver disease diagnosis, distance from a transplant center, and location (rural or urban) of the candidates’ residence were significantly associated with the rate of liver transplantation. Source of insurance payment was the most important factor associated with liver transplantation rates.8 For renal transplant candidates, data suggest that many of these same factors are associated with access to the renal transplant waiting list.9 Evidence also suggests that many potentially appropriate candidates for liver transplantation with alcoholic liver disease are not referred to the transplant system.10 The IOM report also concluded that “the most important predictors of equity of access to transplant services lie outside the organ allocation system—that is, equity to health insurance and high-quality health care services.”11 Thus, although we found small increases in waiting list mortality for the demographic and geographic variables, altering the allocation system is not likely to significantly change access to transplantation because access to transplantation is much more dependent on patients being referred to transplant centers and centers registering candidates on the waiting list, not the allocation of organs once a candidate has been placed on the waiting list. Patients listed at centers with larger waiting lists do not have an increased mortality risk on the waiting list. This indicates that larger centers cannot justify directing more organs to their patients on the basis of the size of their lists. Conversely, we found that patients listed at
centers with smaller lists have increased mortality risk. These centers may be new or relatively inexperienced centers. Centers with relatively small programs may not have the depth of medical and intensive care resources necessary to adequately treat the complex medical problems that patients with liver failure present while waiting. Recent reports have also indicated that patients receiving transplants at smaller centers have increased mortality after transplantation.12 Thus, directing more organs to smaller centers may not necessarily improve the outcome of patients at these smaller centers or the overall efficiency of the allocation system. We found significant variation in the RR for death on the list among centers, among OPOs, and among states of residence. Again, this variation may be the source of considerable controversy and might be used to advocate changes in the allocation system to equalize this variation. That we observed considerable overlap between centers, OPOs, and states with significantly different waiting list mortality risks suggests geographic factors influence waiting list mortality. However, many of these factors are not necessarily influenced by, or likely to be corrected with, changes in the allocation system. Data indicate wide variations in OPO efficiency across the nation and variations in organ donor potential, as well.13 Centers operating alone in a single OPO may have very different listing and organ acceptance criteria compared with centers practicing in OPOs with competitor centers. All these factors are subject to change at any time and probably contribute to the variations in mortality risk we have described. Variations in physician behavior over time make it difficult to quantify the results of organ procurement, allocation, and transplantation and predict results into the future. This is a shortcoming of this analysis and
Liver Transplant Waiting Time and Waiting List Mortality
may make generalization of our findings to present allocation policy difficult because of changes in the medical urgency definitions and sharing agreements. Much more study of these factors must be performed to accurately describe the causes of variations in waiting list mortality among centers, OPOs and states. However, significantly increased risks for waiting list mortality among these geographic areas cannot be ascribed to greater proportions of more medically urgent patients being cared for by centers (or OPOs or states) because we have controlled for medical urgency at the time of entry onto the list in this analysis. Furthermore, whatever the shortcomings of the medical urgency definitions prevailing during our study period, it is clear from our analysis that medical urgency, however defined, has much more association with death on the waiting list than does waiting time. This analysis censored patients still on the list at the close of our study. The fate of these patients could potentially skew the results. However, the majority of centers had median waiting times less than the 2-year follow-up period of our study. For status 3 patients, it is possible that longer follow-up would lead to fewer censored patients, but it is unlikely that this would cause the relationship between mortality and median waiting time to change. Also, the vast majority of patients still waiting at the end of our study period entered the list at status 3 (data not shown). This might explain the inverse trend we saw between waiting list mortality and waiting time. This points to the fallacy of waiting time as a predictor of mortality on the waiting list, especially for patients entering the list at lower medical urgency categories. Also, the absence of minimum listing criteria during our study period probably resulted in wider diversity in the medical urgency of the patients among the UNOS medical urgency categories. However, this lack of uniformity in the UNOS medical urgency categories cannot account for the lack of relationship between the amount of time waiting and risk for dying on the list for all patients entering the waiting list at a similar UNOS medical urgency category, no matter how diverse their medical conditions. Thus, despite the imprecise definitions, our results clearly indicate that the UNOS medical urgency categories as defined during our study have a significant relation to the risk for death on the list. The results of this analysis unmistakably show that during our study period, waiting time had almost no role in determining waiting list mortality. The IOM report also reached essentially the same conclusions on an OPO-specific basis (pretransplantation “mortality rates are relatively constant over time for status 1” and
551
“the longer patients are listed at status 2B or 3, the lower the likelihood of dying on the list”14). Despite the methodologic differences, these 2 separate and independently performed analyses confirm that statusspecific (IOM analysis) and patient-specific (present study) waiting time is not related to the risk for death on the waiting list and should lay to rest the utility of waiting time as a criterion for liver allocation. In addition, the lack of difference in center-specific waiting list mortality for centers performing transplantation on a high proportion of their patients compared with centers performing transplantation on a low proportion of their patients suggests that center-specific transplantation rates also should not influence liver allocation policy. This study only analyzed events before transplantation and did not take into account results after transplantation. There is ample evidence that liver transplantation of more medically urgent patients results in poorer patient and graft survival rates.15,16 Thus, allocation policy must balance the need to minimize waiting list mortality with the desire to maximize transplantation survival. However, it is essential for future policy discussions that valid and reasonable medical outcome measures be used to formulate policy. Data presented here suggest that waiting list mortality is one such criterion that should be considered.
Acknowledgment Members of the UNOS Liver and Intestine Committee, 1997 to 1999: J.G. Turcotte (Chair), University of Michigan; T.K. Howard (Vice Chair), Washington University, St Louis; R.W. Busuttil, University of California at Los Angeles; J.R. Lake, University of Minnesota; R.B. Freeman, Tufts University; J.C. McDonald, Louisiana State University; A. Shaked, University of Pennsylvania; C.W. Pinson, Vanderbilt University; J.S. Bynon, University of Alabama; D.J. Rager, at large; G.A. Halff, University of Texas; J.D. Reyes, University of Pittsburgh; J.P. McVicar, University California at Davis; A.N. Langas, University of Nebraska; K.V. Kowdley, University of Washington; L. Lumeng, Indiana University; M.M.I. Abecassis, Northwestern University; E.S. Maller, Childrens’ of Philadelphia; I. Kam, University of Colorado; E.R. Schiff, University of Miami; M.E. Schwartz, Mt. Sinai School of Medicine; J.W. Springer, at large; J.A. Schulak, University of Cleveland; L.W. Teperman, New York University; P.A. Clavien, Duke University; R.H. Wiesner, Mayo Clinic; H.T. Clark, at large; A.S. Klein, Johns Hopkins University; R.A. Goldstein, Baylor University; J.T. Roberts, University of California at San Francisco; M. Kalayoglu, University of Wisconsin.
552
Freeman and Edwards
References 1. Federal Register 98-8191 (codified at 42 CFR §121), 1998: 16,296. 2. Ubel PA, Caplan AL. Geographic favoritism in liver transplantation—Unfortunate or unfair? N Engl J Med 1998;339:13221325. 3. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Assessing Current Policies and the Potential Impact of the DHHS Final Rule. Washington, DC: National Academy Press; 1999. 4. Krakauer H. Assessing the performance of organ transplant programs: Liver and heart programs. Washington, DC: Office of Health Policy, Office of the Assistant Secretary for Planning and Evaluation, Department of Health and Human Services; 1999. 5. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Washington, DC: National Academy Press; 1999:66. 6. Lucey MR, Brown KA, Everson GT, Fung JJ, Gish R, Keeffe EB, et al. Minimal criteria for placement of adults on the liver transplant waiting list: A report of a national conference organized by the American Society of Transplant Physicians and the American Association for the Study of Liver Diseases. Liver Transpl Surg 1997;3:628-637. 7. Everhart JE, Lombardero M, Detre KM, Zetterman RK, Wiesner RH, Lake JR, Hufnagle JH. Increased waiting time for liver transplantation results in higher mortality. Transplantation 1997;64:1300-1306. 8. Tuttle-Newhall JE, Rutledge R, Johnson M, Fair J. A statewide population-based, time series analysis of access to liver transplantation. Transplantation 1997;63:255-262. 9. Alexander GC, Sehgal AR. Barriers to cadaveric renal transplan-
10. 11.
12.
13.
14.
15.
16.
tation among blacks, women, and the poor. JAMA 1998;280: 1148-1152. Keeffe EB. Assessment of the alcoholic patient for liver transplantation. Liver Transpl Surg 1996;2:12-20. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Assessing Current Policies and the Potential Impact of the DHHS Final Rule. Executive Summary. Washington, DC: National Academy Press; 1999:11. Edwards EB, Roberts JP, McBride MA, Schulak JA, Hunsicker LG. The effect of the volume of procedures at transplantation centers on mortality after liver transplantation. N Engl J Med 1999;341:2049-2053. Luskin RS, Buckley CA, Bradley JW, O’Connor KJ, Delmonico FL. An alternative approach to evaluating organ procurement organization performance. Transplant Proc 1999; 31:353-355. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Assessing Current Policies and the Potential Impact of the DHHS Final Rule. Washington, DC: National Academy Press; 1999:69, 72, 76. Annual Report of the US Scientific Registry for Transplant Recipients and the Organ Procurement and Transplantation Network—Transplant Data: 1988-1997. UNOS, Richmond, VA, and the Division of Transplantation, Bureau of Health Services Administration, US Department of Health and Human Services, Rockville, MD, 1998:139. Muto PA, Freeman RB, Haug CE, Lu A, Rohrer RJ. Liver transplant candidate stratification systems: Implications for third-party payors and organ allocation. Transplantation 1994; 57:306-308
Liver Transplant Waiting Time Does Not Correlate With Waiting List Mortality: Implications for Liver Allocation Policy Richard B. Freeman, Jr,* and Erick B. Edwards,† for the United Network for Organ Sharing Liver and Intestine Committee Factors associated with the risk for mortality once placed on the liver transplant waiting list and how this risk relates to center-specific waiting time and transplant activity have not been adequately evaluated. We performed this study to determine the association between center-specific waiting time and waiting list mortality among liver transplant candidates stratified by medical urgency at the time of registration. A Cox proportional hazards model was used to calculate 2-year mortality risk for a cohort of 16,414 registrants added to the United Network for Organ Sharing liver transplant waiting list between January 1, 1997, and December 31, 1997. After controlling for confounding variables, we calculated the mortality risk for centers, organ procurement organizations (OPOs), and states. The relation between center-specific waiting list mortality risk and median waiting time or transplant activity was determined by linear regression. In multivariate analyses, higher initial medical urgency status (relative risk [RR] ⴝ 12.8; P < .001), increasing age (P < .001), black ethnicity (RR ⴝ 1.29; P < .001), history of previous transplant (RR ⴝ 1.2; P ⴝ .009), certain liver diagnoses, and smaller center size (RR ⴝ 1.39; P ⴝ .008) were associated with significantly increased waiting list mortality. Candidates with blood type A (RR ⴝ 0.87; P < .001) and those with cholestatic cirrhosis as the primary diagnosis (RR ⴝ 0.73; P < 0.001) had a reduced risk for dying. There were significant variations in 2-year waiting list mortality risk among centers, OPOs, and states. However, when stratified by medical urgency status at waiting list entry, center-specific waiting time and transplantation rates accounted for almost none of the center-specific waiting list mortality. Although there are variations in waiting list mortality risk among centers, OPOs, and states, there is very little relation between center-specific waiting list mortality and center-specific median waiting time or center-specific transplantation rates when stratified by medical urgency. Waiting time and center transplant rates should not influence liver allocation policy. (Liver Transpl 2000;6:543-552.)
O
ver the last few years, donor organ allocation policy, and liver allocation in particular, has been intensely scrutinized and criticized in the press and by the federal government.1 Much of the criticism has focused on the wide variations in the time that candidates for liver transplantation must wait for a donor organ and the geographic basis of these differences.2 In answer to these concerns, many liver transplant profes-
sionals have argued that waiting time for a liver transplant is not a good measure of the system’s performance. In 1998, in response to concerns that new proposed federal regulations for national organ allocation policy1 were unduly emphasizing waiting time as a liver allocation measure, Congress commissioned the Institute of Medicine (IOM) to study the liver allocation system.3 In addition, a recent preliminary government study reported wide variations in center-specific liver transplantation rates and implied that this variation might be caused by deficiencies in the allocation system.4 No other analyses of liver transplant waiting list mortality are available in the literature. In designing the present study of liver transplant waiting list outcomes, we chose a relatively uniform cohort of patients to be studied, obtained from a period during which liver allocation policy changed very little. The need to maintain this relative uniformity had to be balanced with the necessity of having an adequate follow-up period to reasonably determine the fate of patients entered onto the waiting list. This study focuses on waiting list mortality as a primary outcome because a more relevant concern is not how long a patient will wait, but what the risk is that a patient will die before getting an opportunity for transplantation. In addition, because patients with liver failure are referred to and managed by transplant centers, not organ procurement organizations (OPOs) or states, we chose to emphasize From the *Department of Surgery, Division of Transplantation, Tufts University School of Medicine/New England Medical Center, Boston, MA; and †United Network for Organ Sharing, Richmond, VA. Presented in part at the 50th Annual Meeting of the American Association for the Study of Liver Diseases, Dallas, TX, November 7, 1999. Address reprint requests to Richard B. Freeman, Jr, MD, Division of Transplantation, New England Medical Center, Box 40, 750 Washington St, Boston, MA 02111. Telephone: 617-636-5592; FAX: 617-6368228; E-mail: [email protected] Copyright © 2000 by the American Association for the Study of Liver Diseases 1527-6465/00/0605-0003$3.00/0 doi:10.1053/jlts.2000.9744
Liver Transplantation, Vol 6, No 5 (September), 2000: pp 543-552
543
544
Freeman and Edwards
center-specific analyses of waiting time and mortality risk. We also chose to define waiting time as the total time a patient spent on the list and stratify our analysis by initial medical urgency status at the time of entry onto the list. This is in contrast to the IOM method in which waiting time was defined as the amount of time a patient spent at a given medical urgency status while that patient was an active candidate.5 This aim of this report is to analyze the risk factors associated with liver transplant waiting list mortality and how center-specific waiting list mortality is related to center-specific waiting time and transplantation activity.
Methods Our study cohort consisted of 16,414 liver transplant candidates with registrations received by the United Network for Organ Sharing (UNOS) between January 1, 1996, and December 31, 1997. We intentionally chose this cohort of patients from a recent 2-year period during which liver allocation policy was stable and relatively current. In addition, this cohort had sufficient follow-up time to accurately determine the risk for death within 2 years of registration. Our primary outcome variable was death within 2 years of registration. Patients who were removed from the waiting list, underwent transplantation, or still waiting at the time of the analysis were censored. Patients removed from the waiting list for being too sick to undergo transplantation were scored as deaths on the list. These patients’ dates of death were randomly assigned to be 0 to 30 days from their date of removal from the waiting list. Seven distinct covariables were examined in a multivariable analysis to determine the relative risk (RR) for waiting list mortality. These were: (1) UNOS medical urgency status at the time of registration, (2) candidates’ blood group, (3) history of previous liver transplant, (4) candidates’ race, (5) age at registration, (6) primary diagnosis at listing, and (7) size of the center listing patients. Center size was defined as the number of new additions to the waiting list entered by each center during the study period. For center-size comparisons, centers were divided into quartiles based on their total new additions to the list. A Cox proportional hazards model was constructed to estimate the RR for waiting list mortality for the various patient and institutional factors previously identified. The proportional hazards assumption was assessed graphically for the statistically significant patient factors. Factors not satisfying this assumption were either dropped from the analysis or included as stratification factors in the Cox model. Once the final set of patient factors was identified, tests of significance were constructed for the institutional factors while simultaneously adjusting for significant patient factors. We did not include patients entered at the lowest medical urgency status (status 4) in our center-specific analyses. This category was eliminated in 1997 because essentially no patients in this
status level underwent transplantation and therefore is irrelevant to current practice. To assess the relation between waiting time and RR for waiting list mortality, we stratified our analyses by initial medical urgency status at the time of entry onto the list. This was done to address the criticisms of previous studies in which no stratification was performed and to comply with the broad opinion of transplant professionals that medical urgency must be taken into account when comparing waiting times to liver transplantation. Median waiting time for patients stratified by initial medical urgency status was determined by calculating the time at which 50% of all patients initially listed at a given medical urgency status received a transplant at each center. In this study, in contrast to the analysis of the IOM, waiting time was patient specific and cumulative. For example, waiting time for patients initially listed at medical urgency status 3 at a given center was calculated from each patient’s initial list date to their corresponding transplantation date regardless of other changes in medical urgency status between the list date and transplantation date. Similarly, for patients initially listed at status 1 or 2, waiting time was calculated from the initial list date to transplantation date, regardless of whether the patient’s medical urgency status changed between the list date and transplantation date. Conversely, the IOM analysis used the time all patients spent at a given medical urgency status to calculate status-specific waiting times regardless of whether a given patient was initially listed at that status. In the IOM calculations, patients who accrued waiting time at a given medical urgency status and then moved to another medical urgency status were censored in the analysis of waiting time for the first status.5 Each center’s patients were then stratified by medical urgency status at the time of listing, and their RR for death within 2 years of registration on the waiting list was calculated. Each center’s median waiting time was plotted against the RR for waiting list mortality for each center’s patients at each medical urgency status to determine the center-specific relation between waiting time and waiting list mortality. We also analyzed the relationship between the percentage of patients who underwent transplantation at each center and center-specific waiting list mortality risk within the 2-year study period. Again stratifying for medical urgency at time of entry onto the list, we plotted the RR for mortality against the percentage of patients listed who received transplants at the respective centers. This was done to assess the relationship between transplant activity and waiting list mortality for each center within each medical urgency category. A simple linear regression was used to determine these relationships. The results are plotted as scatterplots to better show the variation and provide a more complete picture of the data.
Results During our study period, 16,414 candidates were added to the UNOS liver transplant list. Study closeout date was May 9, 1999. All patients were followed up
545
Liver Transplant Waiting Time and Waiting List Mortality
until transplantation, death, or removal from the list, with 3,835 candidates (23.4%) still active on the waiting list at study closeout. One hundred fifteen centers from 52 OPOs added patients to the liver transplant waiting list. Candidates were from all 50 states. During our study period, prevailing organ allocation policy did not include minimum listing criteria,6 and entry onto the list at the lowest medical urgency was at the discretion of the transplant centers. Patients meeting the highest medical urgency status (status 1) were defined as those with acute or chronic liver disease who were confined to an intensive care unit on life support, with less than 1 week to live. Patients meeting the second most urgent status (status 2) were defined as requiring constant hospitalization. Those listed as status 3 required constant medical attention but not necessarily hospitalization, and status 4 included all others. The RRs for mortality once placed on the waiting list stratified by various patient factors are listed in Table 1. Patients entering onto the waiting list at the highest medical urgency (status 1) had more than a 12 times greater risk for death while on the list compared with those entering at the lowest 2 medical urgency
Table 1. Relative Risk for Waiting List Mortality Within 2 Years of Listing According to Patient Characteristics Factor Status at entry 1 2 3 Blood group A B, AB, O Previous OLT Race Black Other White, Asian, Hispanic Age at listing (yr) ⱖ60 v ⱕ45 Diagnosis at listing Cholestatic Laennec’s Noncholestatic Acute Malignancy Unknown Posthepatitic cirrhosis
Relative Risk
P
12.80 5.28 1.00
⬍.001 ⬍.001 Reference
0.87 1.00 1.20
⬍.001 Reference .009
1.29 1.59 1.00
⬍.001 .009 Reference
1.20
⬍.001
0.73 1.17 1.19 1.73 1.91 0.67 1.00
⬍.001 .003 .004 ⬍.001 ⬍.001 .012 Reference
Abbreviation: OLT, orthotopic liver transplantation.
Table 2. Relative Risk for Waiting List Mortality Within 2 Years of Listing According to Center Size Factor First quartile (0 ⬍ additions ⱕ 33) Second quartile (33 ⬍ additions ⱕ 106) Fourth quartile (181 ⬍ additions)
Relative Risk
P
1.39
.008
1.18
.002
1.00
Reference
NOTE. Center size is determined by number of new additions to the waiting list.
categories. Patients entering at medical urgency status 2 had more than a 5 times greater risk for death while on the list compared with those in the lowest 2 medical urgency categories. Patients with blood type A had a reduced risk for death while waiting compared with patients of other blood types. Patients with blood types B, AB, or O did not have significantly different mortality risks when individually compared with all other blood types combined (data not shown). Those with a previous liver transplant, of African descent, or of increasing age had slight but statistically significant increased risks for death on the waiting list. Patients with diagnoses of alcoholic, noncholestatic, or acute liver failure or hepatic malignancy had increased risks for death compared with patients with posthepatitic cirrhosis, whereas patients with cholestatic liver disease had a reduced RR for mortality on the list. New additions to the liver transplant list were tabulated for each liver transplant center during the study period, and the centers were then stratified in quartiles to determine the relation between the size of a liver transplant center’s waiting list and waiting list mortality (Table 2). Centers with the fewest additions to the list had significantly increased RRs for waiting list death compared with centers with the largest waiting lists. The RR for waiting list mortality for each center was adjusted for significant patient factors (waiting list entry status, race, age, blood type, and liver failure diagnosis) and plotted in Figure 1 to show the variation among the centers. Of the 115 centers, 14 centers had statistically significant increased RRs for death (RR range, 1.25 to 4.8) compared with all other centers, and 9 centers had significantly decreased RRs (RR range, 0.31 to 0.73). Figure 2 represents a similar analysis, plotting RR for waiting list mortality against candidates’ state of residence. Patients from 7 states had a statistically increased RR for death on the waiting list (RR range, 1.23 to
546
Freeman and Edwards
Figure 1. Relative risk for pretransplantation candidate mortality within 2 years of listing (vertical axis) shown by center (horizontal axis) for new additions to the liver transplant waiting list, 1996 to 1997. ( ) Centers with relative risks significantly different from all other centers. ID, identifier.
2.37) compared with all other states, and 3 states’ residents had a reduced RR for death (RR range, 0.77 to 0.83). For OPOs, patients listed with centers in 10 OPOs had significantly increased risks for death on the waiting list (RR range, 1.25 to 2.26), and 4 OPOs had a reduced risk for death on the list (RR range, 0.67 to 0.79; Fig. 3). Of the 14 centers with significantly greater mortality rates, 9 centers were in states that had significantly greater mortality rates. Eleven of the 14 centers were associated with OPOs that had significantly greater mortality rates. Of the 9 centers with significantly lower mortality rates, 6 centers were in states with significantly lower mortality rates. Six of the 9 centers were associated with OPOs that had significantly lower mortality rates. In Figure 4, each center’s median waiting time for
Figure 2. Relative risk for pretransplantation candidate mortality within 2 years of listing (vertical axis) shown by candidates’ state of residence (horizontal axis) for new additions to the liver transplant waiting list, 1996 to 1997. ( ) Centers with relative risks significantly different from all other centers. ID, identifier.
patients entering onto the list as status 1 during our study period are plotted against the RR for death on the waiting list for those same center-specific patients. There was no relation between the center-specific median waiting time and the center-specific RR for death for patients initially listed at status 1 (R ⫽ 0.03; R2 ⫽ 0.0009; P ⫽ .84). The few outlier points with extremely long waiting times were caused by a few centers listing only 1 or 2 patients as status 1 during our study period. These patients were downgraded to lower medical urgency levels, making their waiting times extremely long. When these outliers are excluded from the analysis, there is no significant change in the results (data not shown). To further address this phenomenon, we plotted the RR for death on the list for each center’s patients entering onto the list at status 1 against the percentage of those patients who underwent transplantation within
Liver Transplant Waiting Time and Waiting List Mortality
547
Figure 3. Relative risk for pretransplantation candidate mortality within 2 years of listing (vertical axis) shown by OPO (horizontal axis) for new additions to the liver transplant waiting list, 1996 to 1997. ( ) Centers with relative risks significantly different from all other centers. ID, identifier.
our study period at the respective centers (Fig. 5). Again, there is no relation between the center-specific RR for mortality on the list and the center-specific proportion of patients who underwent transplantation who were initially listed at status 1 (R ⫽ 0.033; R2 ⫽ 0.0011; P ⫽ .74). For liver transplant candidates initially entering onto the waiting list at medical urgency status 2, we found very similar results. There is no relation between the center-specific median waiting time and the centerspecific RR for death for patients initially listed at status 2 (Fig. 6; R ⫽ 0.11; R2 ⫽ 0.0117; P ⫽ .47). Again, several outlier points occurred because of the phenomena of a few centers initially listing a very small number of patients at status 2. Elimination of these outliers does not change the results of this correlation (data not shown). There was a very slight trend toward a reduced risk for death for patients listed at centers performing trans-
plantation on a greater proportion of their patients listed at status 2. Although this trend reached statistical significance, less than 5% of the variations in centerspecific mortality risk for status 2 patients can be attributed to the center-specific transplantation rate (Fig. 7; R ⫽ – 0.22; R2 ⫽ 0.0485; P ⫽ .024). For patients initially entering onto the list at status 3 (Fig. 8), there was a trend toward a reduced risk for center-specific waiting list mortality for patients listed at centers with the longer median waiting times (R ⫽ – 0.31; R2 ⫽ 0.0965; P ⫽ .004). This trend indicates a reduced risk for death for the longest waiting status 3 patients when analyzed on a centerspecific basis. For patients initially entering onto the list at status 3, there was no relation between the center-specific RR for mortality on the list and center-specific proportion of patients who underwent transplantation (Fig. 9; R ⫽ 0.08; R2 ⫽ 0.0074; P ⫽ .56).
Figure 4. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 1 (vertical axis) plotted against center-specific median waiting times (horizontal axis).
548
Freeman and Edwards
Figure 5. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 1 (vertical axis) plotted against center-specific percentage of patients listed who received transplants during the study period (horizontal axis).
Discussion Our analysis distinctly indicates several patient factors associated with increased risk for death on the waiting list. By far, the medical condition of the patients at the time of entry onto the list, defined by UNOS medical urgency status, was the most important factor overall associated with increased waiting list mortality. In our analysis, stratification of waiting list candidates by UNOS medical urgency criteria prevailing at the time of our study may not have been the most accurate measure of severity of liver disease. Nonetheless, that both our study and that of the IOM found waiting list mortality to be highly correlated with the UNOS medical urgency criteria indicates that these criteria have validity in predicting waiting list mortality (and therefore urgency for transplantation) for this cohort. The other patient variables associated with increased
Figure 6. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 2 (vertical axis) plotted against center-specific median waiting times (horizontal axis).
mortality risk were several times smaller in magnitude. Patients with a primary liver diagnosis of acute liver failure most often were in the highest medical urgency status, so it is not surprising that these patients also had an increased mortality risk. These data might be used to justify further direction of donor livers to the most medically urgent patients to try to alleviate some of their increased risk for death. However, on further examination, many other factors may be responsible for these findings. Having an available organ at the moment when these patients present to the waiting list is a very random event. Our data showing no more than a trivial relationship between median waiting time for status 1 (or status 2) patients and a negative relationship for status 3 patients and waiting list mortality support this notion. One might conclude that centers with longer median waiting times for status 1 patients might have higher (or lower) waiting list mortality risk if pa-
Liver Transplant Waiting Time and Waiting List Mortality
549
Figure 7. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 2 (vertical axis) plotted against center-specific percentage of patients listed who received transplants during the study period (horizontal axis).
tient disease progression and organ availability were both constant linear phenomena. Conversely, we found virtually no relation between waiting time and mortality for each medical urgency status, indicating that time on the list is not associated with an increased or decreased risk for death without a transplant. In addition, patients listed at centers performing transplantation on a high percentage of candidates initially listed at status 1 or 3 did not have a significantly increased (or decreased) risk for mortality on the list, and the variation in waiting list mortality risk attributable to center transplantation rates for status 2 patients was less than 5%. Therefore, our data do not directly support even more priority being given to these more critically ill patients to reduce their waiting time than was prevailing during our study period. That the other patient factors associated with increased risk for death on the list are relatively small
makes it difficult to advocate changes in the allocation system to equalize these risks. The effect of blood type on pre–liver transplantation mortality has been examined, with a slight but significant increase in pretransplantation mortality for candidates with blood type O compared with all other blood types.7 This report also found patients with blood type O waited longer for liver transplants and concluded that waiting time was associated with mortality. However, the data for this previous study came from only 3 large liver transplant centers and was culled from 1990 to 1993, when waiting times were dramatically shorter (median waiting time, 109 days for blood type O, 58 days for blood types A, B, and AB). Also, this analysis was not performed on a center-specific basis. The use of blood type as a surrogate for randomization to transplantation after longer or shorter waiting times also has limitations, as noted by the investigators of this study.
Figure 8. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 3 (vertical axis) plotted against center-specific median waiting times (horizontal axis).
550
Freeman and Edwards
Figure 9. Center-specific relative risk for waiting list mortality for liver patients initially listed at status 3 (vertical axis) plotted against center-specific percentage of patients listed who received transplants during the study period (horizontal axis).
Factors such as age, race, and primary liver diagnosis are as much a function of referral patterns, insurance contracts, and demographics of referral areas as they are likely to be influenced by the allocation system. In the 1 report examining access to liver transplantation before entry onto the allocation list, demographic factors of age, sex, source of medical insurance payment, liver disease diagnosis, distance from a transplant center, and location (rural or urban) of the candidates’ residence were significantly associated with the rate of liver transplantation. Source of insurance payment was the most important factor associated with liver transplantation rates.8 For renal transplant candidates, data suggest that many of these same factors are associated with access to the renal transplant waiting list.9 Evidence also suggests that many potentially appropriate candidates for liver transplantation with alcoholic liver disease are not referred to the transplant system.10 The IOM report also concluded that “the most important predictors of equity of access to transplant services lie outside the organ allocation system—that is, equity to health insurance and high-quality health care services.”11 Thus, although we found small increases in waiting list mortality for the demographic and geographic variables, altering the allocation system is not likely to significantly change access to transplantation because access to transplantation is much more dependent on patients being referred to transplant centers and centers registering candidates on the waiting list, not the allocation of organs once a candidate has been placed on the waiting list. Patients listed at centers with larger waiting lists do not have an increased mortality risk on the waiting list. This indicates that larger centers cannot justify directing more organs to their patients on the basis of the size of their lists. Conversely, we found that patients listed at
centers with smaller lists have increased mortality risk. These centers may be new or relatively inexperienced centers. Centers with relatively small programs may not have the depth of medical and intensive care resources necessary to adequately treat the complex medical problems that patients with liver failure present while waiting. Recent reports have also indicated that patients receiving transplants at smaller centers have increased mortality after transplantation.12 Thus, directing more organs to smaller centers may not necessarily improve the outcome of patients at these smaller centers or the overall efficiency of the allocation system. We found significant variation in the RR for death on the list among centers, among OPOs, and among states of residence. Again, this variation may be the source of considerable controversy and might be used to advocate changes in the allocation system to equalize this variation. That we observed considerable overlap between centers, OPOs, and states with significantly different waiting list mortality risks suggests geographic factors influence waiting list mortality. However, many of these factors are not necessarily influenced by, or likely to be corrected with, changes in the allocation system. Data indicate wide variations in OPO efficiency across the nation and variations in organ donor potential, as well.13 Centers operating alone in a single OPO may have very different listing and organ acceptance criteria compared with centers practicing in OPOs with competitor centers. All these factors are subject to change at any time and probably contribute to the variations in mortality risk we have described. Variations in physician behavior over time make it difficult to quantify the results of organ procurement, allocation, and transplantation and predict results into the future. This is a shortcoming of this analysis and
Liver Transplant Waiting Time and Waiting List Mortality
may make generalization of our findings to present allocation policy difficult because of changes in the medical urgency definitions and sharing agreements. Much more study of these factors must be performed to accurately describe the causes of variations in waiting list mortality among centers, OPOs and states. However, significantly increased risks for waiting list mortality among these geographic areas cannot be ascribed to greater proportions of more medically urgent patients being cared for by centers (or OPOs or states) because we have controlled for medical urgency at the time of entry onto the list in this analysis. Furthermore, whatever the shortcomings of the medical urgency definitions prevailing during our study period, it is clear from our analysis that medical urgency, however defined, has much more association with death on the waiting list than does waiting time. This analysis censored patients still on the list at the close of our study. The fate of these patients could potentially skew the results. However, the majority of centers had median waiting times less than the 2-year follow-up period of our study. For status 3 patients, it is possible that longer follow-up would lead to fewer censored patients, but it is unlikely that this would cause the relationship between mortality and median waiting time to change. Also, the vast majority of patients still waiting at the end of our study period entered the list at status 3 (data not shown). This might explain the inverse trend we saw between waiting list mortality and waiting time. This points to the fallacy of waiting time as a predictor of mortality on the waiting list, especially for patients entering the list at lower medical urgency categories. Also, the absence of minimum listing criteria during our study period probably resulted in wider diversity in the medical urgency of the patients among the UNOS medical urgency categories. However, this lack of uniformity in the UNOS medical urgency categories cannot account for the lack of relationship between the amount of time waiting and risk for dying on the list for all patients entering the waiting list at a similar UNOS medical urgency category, no matter how diverse their medical conditions. Thus, despite the imprecise definitions, our results clearly indicate that the UNOS medical urgency categories as defined during our study have a significant relation to the risk for death on the list. The results of this analysis unmistakably show that during our study period, waiting time had almost no role in determining waiting list mortality. The IOM report also reached essentially the same conclusions on an OPO-specific basis (pretransplantation “mortality rates are relatively constant over time for status 1” and
551
“the longer patients are listed at status 2B or 3, the lower the likelihood of dying on the list”14). Despite the methodologic differences, these 2 separate and independently performed analyses confirm that statusspecific (IOM analysis) and patient-specific (present study) waiting time is not related to the risk for death on the waiting list and should lay to rest the utility of waiting time as a criterion for liver allocation. In addition, the lack of difference in center-specific waiting list mortality for centers performing transplantation on a high proportion of their patients compared with centers performing transplantation on a low proportion of their patients suggests that center-specific transplantation rates also should not influence liver allocation policy. This study only analyzed events before transplantation and did not take into account results after transplantation. There is ample evidence that liver transplantation of more medically urgent patients results in poorer patient and graft survival rates.15,16 Thus, allocation policy must balance the need to minimize waiting list mortality with the desire to maximize transplantation survival. However, it is essential for future policy discussions that valid and reasonable medical outcome measures be used to formulate policy. Data presented here suggest that waiting list mortality is one such criterion that should be considered.
Acknowledgment Members of the UNOS Liver and Intestine Committee, 1997 to 1999: J.G. Turcotte (Chair), University of Michigan; T.K. Howard (Vice Chair), Washington University, St Louis; R.W. Busuttil, University of California at Los Angeles; J.R. Lake, University of Minnesota; R.B. Freeman, Tufts University; J.C. McDonald, Louisiana State University; A. Shaked, University of Pennsylvania; C.W. Pinson, Vanderbilt University; J.S. Bynon, University of Alabama; D.J. Rager, at large; G.A. Halff, University of Texas; J.D. Reyes, University of Pittsburgh; J.P. McVicar, University California at Davis; A.N. Langas, University of Nebraska; K.V. Kowdley, University of Washington; L. Lumeng, Indiana University; M.M.I. Abecassis, Northwestern University; E.S. Maller, Childrens’ of Philadelphia; I. Kam, University of Colorado; E.R. Schiff, University of Miami; M.E. Schwartz, Mt. Sinai School of Medicine; J.W. Springer, at large; J.A. Schulak, University of Cleveland; L.W. Teperman, New York University; P.A. Clavien, Duke University; R.H. Wiesner, Mayo Clinic; H.T. Clark, at large; A.S. Klein, Johns Hopkins University; R.A. Goldstein, Baylor University; J.T. Roberts, University of California at San Francisco; M. Kalayoglu, University of Wisconsin.
552
Freeman and Edwards
References 1. Federal Register 98-8191 (codified at 42 CFR §121), 1998: 16,296. 2. Ubel PA, Caplan AL. Geographic favoritism in liver transplantation—Unfortunate or unfair? N Engl J Med 1998;339:13221325. 3. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Assessing Current Policies and the Potential Impact of the DHHS Final Rule. Washington, DC: National Academy Press; 1999. 4. Krakauer H. Assessing the performance of organ transplant programs: Liver and heart programs. Washington, DC: Office of Health Policy, Office of the Assistant Secretary for Planning and Evaluation, Department of Health and Human Services; 1999. 5. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Washington, DC: National Academy Press; 1999:66. 6. Lucey MR, Brown KA, Everson GT, Fung JJ, Gish R, Keeffe EB, et al. Minimal criteria for placement of adults on the liver transplant waiting list: A report of a national conference organized by the American Society of Transplant Physicians and the American Association for the Study of Liver Diseases. Liver Transpl Surg 1997;3:628-637. 7. Everhart JE, Lombardero M, Detre KM, Zetterman RK, Wiesner RH, Lake JR, Hufnagle JH. Increased waiting time for liver transplantation results in higher mortality. Transplantation 1997;64:1300-1306. 8. Tuttle-Newhall JE, Rutledge R, Johnson M, Fair J. A statewide population-based, time series analysis of access to liver transplantation. Transplantation 1997;63:255-262. 9. Alexander GC, Sehgal AR. Barriers to cadaveric renal transplan-
10. 11.
12.
13.
14.
15.
16.
tation among blacks, women, and the poor. JAMA 1998;280: 1148-1152. Keeffe EB. Assessment of the alcoholic patient for liver transplantation. Liver Transpl Surg 1996;2:12-20. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Assessing Current Policies and the Potential Impact of the DHHS Final Rule. Executive Summary. Washington, DC: National Academy Press; 1999:11. Edwards EB, Roberts JP, McBride MA, Schulak JA, Hunsicker LG. The effect of the volume of procedures at transplantation centers on mortality after liver transplantation. N Engl J Med 1999;341:2049-2053. Luskin RS, Buckley CA, Bradley JW, O’Connor KJ, Delmonico FL. An alternative approach to evaluating organ procurement organization performance. Transplant Proc 1999; 31:353-355. Committee on Organ Procurement and Transplantation Policy, Institute of Medicine. Organ Procurement and Transplantation. Assessing Current Policies and the Potential Impact of the DHHS Final Rule. Washington, DC: National Academy Press; 1999:69, 72, 76. Annual Report of the US Scientific Registry for Transplant Recipients and the Organ Procurement and Transplantation Network—Transplant Data: 1988-1997. UNOS, Richmond, VA, and the Division of Transplantation, Bureau of Health Services Administration, US Department of Health and Human Services, Rockville, MD, 1998:139. Muto PA, Freeman RB, Haug CE, Lu A, Rohrer RJ. Liver transplant candidate stratification systems: Implications for third-party payors and organ allocation. Transplantation 1994; 57:306-308