[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Download PDF
Image description not available.
Percentages of programs with actual graft survival rates above, below, or equal to expected survival rates.
Table 1.—Number of Transplants and Transplantation Programs
Image description not available.
Table 2.—Completeness of Follow-up*
Image description not available.
Table 3.—Graft and Patient Survival*
Image description not available.
Table 4.—One-Year Graft and Patient Survival by Era*
Image description not available.
Table 5.—Donor and Recipient Factors Affecting 1-Year and Conditional 3-Year Kidney Graft Survival*
Image description not available.
Table 6.—Donor and Recipient Factors Affecting 1-Year and Conditional 3-Year Liver Survival*
Image description not available.
Table 7.—Donor and Recipient Factors Affecting 1-Year and Conditional 3-Year Heart Graft Survival*
Image description not available.
1.
Edwards E. 1991 Center Specific Graft and Patient Survival Rates . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1992.
2.
Edwards E, Bennett L, Guo T.  et al.  1994 Center Specific Graft and Patient Survival Rates . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1994.
3.
Burdick J, Norman DJ, Hunsicker L.  et al.  Identification of poorly performing transplant centers using the UNOS center-specific data.  Transplant Proc.1997;29:1495.
4.
Smith CM, Ellison MD, White R.  et al.  1996 Annual Report of the US Scientific Registry for Transplant Recipients and the Organ Procurement and Transplantation Network . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1996.
5.
Lin H-M, McBride MA, Davies D.  et al.  1997 Center Specific Graft and Patient Survival Rates . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1997.
6.
Copeland GP, Jones D, Walters M. POSSUM: a scoring system for surgical audit.  Br J Surg.1991;78:355-360.
7.
Grover FL, Hammermeister KE, Burchfiel C.  et al.  Initial report for the Veterans Administration preoperative risk assessment study for cardiac surgery.  Ann Thorac Surg.1990;50:12-28.
8.
Edwards FH, Clark RE, Schwartz M. Coronary artery bypass grafting.  Ann Thorac Surg.1994;57:12-19.
9.
Khuri SF, Dailey J, Henderson W.  et al.  Risk adjustment of the postoperative mortality rate for the comparative assessment of the quality of surgical care: results of the National Veterans Affairs Surgical Risk Study.  J Am Coll Surg.1997;185:315-327.
10.
Hannan EL, Kilburn H, Racz M.  et al.  Improving the outcomes of coronary artery bypass surgery in New York State.  JAMA.1994;271:761-766.
11.
Hannan EL, Siu AL, Kumar D.  et al.  The decline in coronary artery bypass graft surgery mortality in New York State: the role of surgeon volume.  JAMA.1995;273:209-213.
12.
O'Connor GT, Plume SK, Olmstead EM.  et al.  A regional intervention to improve the hospital mortality associated with coronary artery bypass surgery.  JAMA.1996;275:841-846.
13.
Wennberg JE, Freeman JL, Shelton RM.  et al.  Hospital use and mortality among Medicare beneficiaries in Boston and New Haven.  N Engl J Med.1989;321:1168-1173.
14.
Kauffman HM, Bennett LE, McBride MA, Ellison MD. The expanded donor.  Transplant Rev.1997;11:165-190.
15.
Katznelson S, Cecka JM. Immunosuppressive regimens and their effects on renal allograft outcome. In: Cecka JM, Terasaki PI, eds. Clinical Kidney Transplants, 1996 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1997:361.
16.
Klintmalm G. A review of FK506: a new immunosuppressant agent for the prevention and rescue of graft rejection.  Transplant Rev.1994;8:53-63.
17.
Opelz G, Mickey MR, Terasaki PI. Comparison of kidney transplant survival among transplant centers.  Transplantation.1975;19:226-229.
18.
Mickey MR. Center variability. In: Terasaki PI, ed. Clinical Kidney Transplants, 1989 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1989:435-446.
19.
Cicciarelli J. Transplant center and kidney graft survival. In: Terasaki PI, ed. Clinical Kidney Transplants, 1985 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1986:93.
20.
Cho YW, Cecka JM. Center effect in the UNOS renal transplant registry. In: Terasaki PI, ed. Clinical Kidney Transplants, 1992 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1993:333.
21.
Cho YW, Cecka JM. Organ procurement organization and transplant center effects on cadaver renal outcomes. In: Terasaki PI, ed. Clinical Kidney Transplants, 1996 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1997:427.
22.
Khuri SF, Daley J, Henderson W.  et al.  The national Veterans Administration Surgical Risk Study.  J Am Coll Surg.1995;180:519-531.
23.
Le A, Wilson R, Douek K. Prospective risk stratification in renal transplant candidates for cardiac death.  Am J Kidney Dis.1994;24:65-71.
24.
Butkus DE, Meydrech EF, Raju SS. Racial differences in the survival of cadaveric renal allografts: overriding effects of HLA matching and socioeconomic factors.  N Engl J Med.1992;327:840-845.
25.
Cecka JM, Gjertson D, Terasaki PI. Do prophylactic antilymphocyte globulins (ALG and OKT3) improve renal transplant survival in recipient and donor high-risk groups?  Transplant Proc.1993;25:548-549.
26.
Shield CF, Edwards EB, Davies DB, Daily OP. Antilymphocyte induction therapy in cadaver renal transplantation.  Transplantation.1997;63:1257-1263.
27.
Hosenpud JD, Breen TJ, Edwards EB.  et al.  The effect of transplant center volume on cardiac transplant outcome.  JAMA.1994;271:1844-1849.
28.
Hannan EL, O'Donnell JF, Kilburn H.  et al.  Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals.  JAMA.1989;262:503-510.
Original Contribution
October 7, 1998

Center-Specific Graft and Patient Survival Rates1997 United Network for Organ Sharing (UNOS) Report

Author Affiliations

From the United Network for Organ Sharing, Richmond, Va (Drs Lin, Kauffman, McBride, Edwards, and Daily, Mmes Davies and Smith, and Mr Rosendale); the Department of Health Evaluation Sciences, Milton S. Hershey Medical Center, Hershey, Pa (Dr Lin); Cardiac Surgery Division, University of Alabama Hospital, Birmingham (Dr Kirklin); Department of Surgery, Via Christi Regional Medical Center, Wichita, Kan (Dr Shield); and the Department of Internal Medicine, University of Iowa Hospitals and Clinics, Iowa City (Dr Hunsicker).

JAMA. 1998;280(13):1153-1160. doi:10.1001/jama.280.13.1153
Context.—

Context.— Multiple comprehensive, risk-adjusted studies evaluating short-term surgical mortality have been reported previously. This report analyzes short-term and long-term outcomes, both nationally and at each individual transplant program, for all solid organ transplantations performed in the United States.

Objectives.— To report graft and patient survival rates for all solid organ transplantations, both nationally and at each specific transplant program in the United States, and to compare the expected survival rate with the actual survival rate of each individual program.

Design and Setting.— Multivariate regression analysis of donor and recipient factors affecting graft and patient survival of all kidney, liver, pancreas, heart, lung, and heart-lung transplants reported to the United Network for Organ Sharing from 742 separate transplant programs.

Patients.— A cohort of 97587 solid organ transplantations performed on 92966 recipients in the United States from January 1988 through April 1994.

Main Outcome Measures.— Short-term and conditional 3-year national and individual transplant program graft and patient survival rates overall and from 2 separate eras (era 1, January 1988-April 1992; era 2, May 1992-April 1994); comparison of actual center-specific performance with risk-adjusted expected performance and identification of centers with better-than-expected or worse-than-expected survival rates.

Results.— One-year graft follow-up exceeded 98% and conditional 3-year follow-up exceeded 91% for all organs. Graft and patient survival improved significantly in era 2 compared with era 1 for all cadaver organs except heart, which remained the same. One-year cadaveric graft survival ranged from 81.5% for heart to 61.9% for heart-lung and 3-year conditional graft survival ranged from 91.3% for pancreas to 74.7% for lung. The percentage of programs whose actual 1-year graft survival was not different from or was better than their risk-adjusted expected survival ranged from 98.3% for heart-lung to 75.7% for liver. Most kidney, liver, and heart programs whose actual survival was significantly less than expected performed small numbers (less than the national average) of transplantations per year.

Conclusions.— Graft and patient survival for solid organ transplantations showed improvement over time. Conditional 3-year graft and patient survival rates were approximately 90% for all organs except for lung and heart-lung. The conditional 3-year survival rates were better than 1-year survival rates, indicating the major risk after transplantation occurs in the first year. The majority of transplant programs achieved actual survival rates not significantly different from their expected survival rates. Center effects were most significant within the first year after transplantation and had much less influence on long-term survival outcomes.

THE UNITED NETWORK for Organ Sharing (UNOS) operates the national Organ Procurement Transplant Network and the Scientific Registry of transplant recipients. Under the Transplant Amendment Act of 1990, UNOS is required by the government to report center-specific transplant survival rates for all solid organ transplantations performed within the United States. The 1991 report1 was based on all solid organ transplantations (N=28858) performed between October 1, 1987, and December 31, 1989, and the 1994 report2 was based on 60100 transplants performed on 57457 patients between October 1, 1987, and December 31, 1991. The 1997 report is the third center-specific report and is based on a cohort of 97587 transplants performed on 92966 recipients from January 1, 1988, through April 30, 1994. The 1997 report contains an executive summary and 6 organ-specific volumes ("Kidney," "Liver," "Pancreas," "Heart," "Lung," and "Heart-Lung").

Major donor and recipient risk factors (negative and positive) affecting national graft and patient outcomes were identified for each organ type using logistic regression. Based on the particular combination of risk factors present within a transplant program's donor and recipient population, expected graft and patient survival rates for each transplantation and, subsequently, for each transplant program were calculated. Then, the actual graft and patient survival rates of each individual program were also determined and then compared with the expected survival rates and analyzed for statistical significance. The results of this analysis determined to which of the 3 groups each transplant program belonged: (1) actual survival rates equaled expected survival rates; (2) actual survival rates exceeded expected survival rates; and (3) actual survival rates were less than expected rates.

Prompted by the results of the 1991 report and the 1994 report, the UNOS Membership and Professional Standards Committee and the UNOS staff developed objective criteria for determining which programs should be further investigated due to lower-than-expected actual survival rates. If the performance of these programs continues to fall below the criteria developed by the committee, the outlier programs will be individually reviewed by designees of the committee. More detail of the review process is described in the "Comment" section and in the article published by Burdick et al.3

METHODS
Participating Organizations

The report consists of data from 275 transplantation centers with 742 separate transplant programs, 66 organ procurement organizations (OPOs), and 56 histocompatibility laboratories.

Data Collection and Validation

UNOS policy requires that all organ transplantations be reported to the UNOS Scientific Registry. Baseline demographic and functional data are collected on all patients at the time they are added to the national waiting list. Further data are reported at the time of transplantation and at the initial discharge from the hospital after the transplant. In addition, data are reported at the time of graft failure, at death, or at 6 months after transplantation for abdominal organs and annually thereafter for all organs. Donor data are submitted by the OPOs, and donor and recipient histocompatibility data are submitted by the respective histocompatibility laboratories. UNOS's data collection procedures and data audits are described in each year's UNOS annual report.4

In the 1997 center-specific report, data were analyzed on all solid, orthotopic organ transplantations performed in the United States from January 1988 through April 1994. Except for kidney, live donor (LD) transplants were excluded. Multiorgan transplants (except heart-lung and kidney-pancreas) also were not included in the report.

After a preliminary data analysis, each transplant program received a complete copy of its own data for review. If the transplant program made corrections on the cadaveric donor or histocompatibility data, corresponding OPOs and histocompatibility laboratories were asked to verify these corrections. The final survival results for each transplant program in this report were based on these validated data. The change rates from prevalidation to postvalidation declined below 3% for the majority of donor and recipient demographics.

Data Analysis

In this report, short-term survival is defined as survival at 3 months (1 month for thoracic organs) and 1 year. Long-term survival is defined in 2 ways as (1) survival at 3 years after transplantation and (2) survival at 3 years after transplantation only for those grafts or patients surviving at least 1 year, hereafter referred to as conditional 3-year survival. The emphasis in the 1997 report is on conditional 3-year survival because it provides an assessment of risk factors independent of those that pose a risk only during the first year, such as the risk of prolonged organ preservation, the surgical procedure itself, and early acute infection and rejection events. The analysis, therefore, is different from the previous center-specific reports in that it evaluates the short-term and the long-term risk factors separately. A subset of the transplant cohort (January 1988-April 1992) was used to determine the long-term survival rates. For short-term survival, cohorts from 2 eras (January 1988-April 1992 and May 1992-April 1994) were analyzed separately to identify any differences in short-term graft and patient survival rates between the 2 eras and assess the performance of each transplant program over time.

Actual graft and patient survival rates were calculated as follows. The survival time for each graft was calculated as the number of days from the transplantation date to the date of graft failure. If graft failure was not reported, the date of death or the date of follow-up (if death was not reported) was used. Furthermore, 1 year after transplantation, death with a functioning kidney or pancreas was censored at the time of death. Patient survival time was calculated as the number of days from transplantation date to death date or, if death was not reported, to the date of follow-up. Actual survival rates were not adjusted for donor and recipient risk factors. Grafts and patients still surviving at specified time points were given a weight of 1. Grafts and patients with incomplete follow-up data received a fractional weight equal to the fraction of time observed. The actual survival rate for a transplant program at a given time point was simply the weighted percentage of grafts or patients that were still functioning or alive at that time point. This method yielded survival rates nearly identical to the survival rates estimated by the Kaplan-Meier method, because follow-up was complete for 98.1% of the transplants in this study.

Expected graft and patient survival rates were estimated at specific time points (1 or 3 months, 1 year, and 3 years conditional on 1-year survival) using a logistic regression model that adjusted for each donor and recipient risk factor. Because the odds ratios of graft and patient survival differ among organ types and are not constant over time, short-term and long-term survival were modeled separately by organ type. For each organ, a set of risk factors was first recommended by an Ad Hoc Technical Oversight Subcommittee of the UNOS Scientific Advisory Committee. Covariates that demonstrated statistical significance (P<.05 for kidney and liver, P<.10 for heart and lung, and P<.15 for heart-lung and pancreas) in the preliminary analysis were then used in the final multivariate analyses (significance levels were determined based on the sample size for each organ type). The final number of donor covariates ranged from 3 (pancreas, lung, heart-lung) to 15 (kidney). The final number of recipient covariates ranged from 6 (heart-lung) to 22 (kidney). As with the actual survival rates, the expected survival rates for a transplant program, at a given time point, were computed as the weighted average of the individual expected survival rates of the patients within the program. Therefore, the expected survival rate for a transplant program represented the average national survival rate for transplants that had the same features as those transplantations performed at that program. Details of the technical methods are described in the "Executive Summary" volume of the 1997 report.5

In the formal 1997 center-specific report, the C statistic is used as a measurement of how well actual survival rates are explained by each logistic regression model.5 Additionally, a modified R2 was calculated to measure how much of the program-to-program variation in actual survival rates is explained by the individual organ models and is expressed as a percentage.

Finally, actual graft and patient survival rates for each individual transplant program were compared with the expected survival rates for that program based on the case mix of risk factors for that program. Based on the statistical significance of the difference of the score statistic5 (P<.05 considered as significant) between the actual survival rate and the expected survival rate, programs were then categorized into one of the following groups: (1) actual survival equals expected survival (P≥0.05); (2) actual survival greater than expected survival (P<.05); or (3) actual survival less than expected survival (P<.05).

RESULTS
Demographic Factors and Follow-up Data

A total of 97587 transplantations in 92966 recipients performed at 742 organ transplant programs were included in the 1997 center-specific report (Table 1). This represented an increase of 37487 transplants over the 60100 recorded in the 1994 report.2 The largest percentage increases since the 1994 report were observed among lung (189%), pancreas (88%), and liver (73%) transplantations.

The number of transplant programs in the 1997 report (742) was 16% greater than the number in the 1994 report (640) and 40% greater than the 1991 report (531). The largest percentage increases since the 1991 report were seen in lung (213%), pancreas (92%), and liver (47%) programs.

Completeness of graft and patient follow-up for 1-year and conditional 3-year survival is shown in Table 2. One-year graft follow-up exceeded 98% for all organs, 1-year patient follow-up exceeded 96% for all organs, and conditional 3-year follow-up for both graft and patient survival exceeded 91% for all organs.

Graft and Patient Survival

Actual graft and patient survival rates at 3 months (1 month for thoracic organs), 1 year, 3 years, and conditional 3 years are shown in Table 3. The short-term survival analyses involved the entire cohort of 97587 transplantations, while the long-term survival analyses only included those transplantations performed from January 1988 through April 1992 (n=62569, in order to allow for long-term follow-up). One-year graft survival rates ranged from a high of 91% for LD kidney transplantations to a low of 62% for heart-lung transplantations. Three-year graft survival rates ranged from a high of 86% for LD kidney transplantations to a low of 50% for heart-lung. However, conditional 3-year graft survival rates were approximately 90% for all organs, with the exception of lung and heart-lung.

In general, abdominal organ transplantations (kidney, pancreas, and liver) had higher patient survival than graft survival due in part to retransplantation for graft failure or return to maintenance therapy (dialysis or exogenous insulin). Retransplantation rates for both kidney and liver were 13%. Pancreas retransplantation rates were low (3.9%), most likely because of the option of insulin injection.

Thoracic organ transplantations had comparable graft and patient survival rates at all time points. Graft failure rates in thoracic organs were similar to death rates, primarily because of rare opportunity for retransplantation when grafts failed (retransplantation rates were <3% for all thoracic organs).

One-year graft and patient survival in 2 separate eras are shown in Table 4. The table shows a substantial improvement in graft and patient survival rates over time for nearly all organs. The only exceptions were LD kidney transplantations and heart transplantations, which remained virtually unchanged in the 2 eras. The greatest improvements in survival were seen for heart-lung, lung, and liver transplants with 1-year graft survival rate increases of 12, 7, and 7 percentage points, respectively, and with 1-year patient survival rate increases of 11, 6, and 5 percentage points, respectively. Factors accounting for the improvement in graft and patient survival from era 1 to era 2 were not examined formally in this study.

Risk Factors Affecting Transplant Survival

In the complete 1997 Report of Center Specific Graft and Patient Survival Rates, tables listing donor and recipient risk factors significantly affecting 1-month (3-month), 1-year, and conditional 3-year graft and patient survival for each type of organ transplantation are presented.5 Statistically significant risk factors affecting kidney, liver, and heart graft survival are reported herein.

Kidney Graft Survival

Donor and recipient risk factors affecting 1-year and conditional 3-year kidney graft survival are shown in Table 5. The noncontinuous donor factors identified as affecting 1-year graft survival were living vs cadaver donor, donor cause of death, donor race, donor sex, and degree of HLA mismatch. Continuous donor factors included cold ischemia time and donor age. For the first 10-hour increase in cold ischemia time (from 23-33 hours), there was a 6% increase in the odds of 1-year graft failure. The mean cadaveric donor age was 32 years. Because of the more substantial effect of donor age at the extremes, it was necessary to model donor age with a quadratic term. The odds ratio for the first 10-year increment from the mean was 1.096 and for the first 20-year increment from the mean was 1.361. This indicates that for a 42-year-old cadaveric donor, there was a 10% increase in the odds of 1-year graft failure, when controlling for all other factors. Similarly, for a 52-year-old cadaveric donor, there was a 36% increase in the odds of 1-year graft failure.

Noncontinuous recipient factors significantly affecting 1-year graft survival were as follows: (1) whether the graft was a primary or repeat transplant; (2) recipient race; (3) whether the graft was a solitary kidney or combined kidney-pancreas transplant; (4) cause of the primary kidney disease; (5) whether the recipient had received any pretransplantation transfusions; (6) whether the recipient was an outpatient, hospitalized, or in an intensive care unit; and (7) the year of transplantation. Continuous recipient factors significantly affecting 1-year graft survival included age, body mass index (calculated as weight in kilograms divided by the square of height in meters), and the peak panel reactive antibody (PRA). As was necessary for the donor age, recipient age required modeling with a quadratic term.

Certain donor factors, such as living vs cadaver donor, black race, donor cause of death, donor age, and degree of HLA mismatch were significant in both the short-term and long-term models. All recipient risk factors that affected short-term survival also significantly affected the conditional 3-year survival, except for recipients younger than 2 years, the severity of illness (ie, hospitalized, intensive care unit, life support), pretransplantation transfusions, and cold ischemia time. The observed reduced odds of graft failure at conditional 3 years for older recipients were not anticipated but might be due to a relatively small and selected sample of older recipients who survived the first year after transplantation. This selection occurred because there was an increased incidence of death with functioning grafts among older recipients. Other factors possibly influencing this observation might be more careful selection of older patients as potential recipients and a decreased incidence of rejection episodes among these older patients. All of the above factors were either donor or recipient characteristics or events that occurred at the time of transplantation. Two events that occurred after the transplants were the need for dialysis in the first week and rejection episodes in the first 6 months. Both were evaluated only in the long-term model. (These 2 covariates, nevertheless, were not used in calculating the long-term center-specific results, because they could have confounded the measurements of the center effect.) After transplantation dialysis increased the odds of graft failure by 21% and the presence of any posttransplant rejection episode increased the odds of graft failure by 49%.

Liver Graft Survival

Donor and recipient factors affecting short-term and long-term liver graft survival are shown in Table 6. For 1-year graft survival, significant noncontinuous donor factors were donor cause of death (eg, cerebrovascular event such as stroke) and donor race and sex (black, Hispanic, and female). Use of reduced size or split liver grafts as well as ABO blood type incompatible organs also significantly affected the odds of survival. Continuous donor factors included donor age and cold ischemia time. The mean cadaveric liver donor age was 28 years and donor age was modeled using a quadratic term. For the first 10-year increment from the mean donor age, there was a 13% increase in 1-year graft failure; for the next 10-year increment, there was an additional 17% increase in 1-year graft failure. Similarly, the mean ischemia time for livers was 11 hours and for each 10-hour increase in cold ischemia time, there was a 12% increase in the odds of 1-year graft failure.

Significant recipient risk factors included whether the patient was on any form of life support (usually a ventilator) or was hospitalized or in an intensive care unit. When the recipient's preoperative serum creatinine level was greater than 180 µmol/L (2 mg/dL), the odds ratio for graft failure increased to 1.50. The odds ratios when the primary disease process was hepatic malignancy or when the patient was undergoing a repeat transplantation were 1.65. Graft survival for Asian recipients was significantly less than that for other racial groups. Recipient age also was shown to affect graft outcome; recipients younger than 1 year exhibited poorer survival. The mean recipient age was 40 years, and recipient age was modeled with a quadratic term. For the first 10-year increment from the mean age, there was an 11% increase in 1-year graft failure; for the next 10-year increment, there was an additional 21% increase in 1-year graft failure. Overall graft survival improved over time as transplantations performed in 1993 to 1994 had a 41% reduced odds of graft failure over transplantations performed in 1988 to 1989.

In the conditional 3-year model, there were significantly increased risks of long-term graft failure among recipients of reduced and split livers, repeat transplant, recipients younger than 1 year, and black recipients. Patients diagnosed as having malignant neoplasms had nearly 6 times the odds of long-term graft failure when compared to the cirrhosis disease group. In contrast, recipients with acute hepatic necrosis, cholestatic disease, or biliary cirrhosis had a 33% to 60% lower odds of long-term graft failure. Medical urgency indicators, such as being in the intensive care unit or receiving life support and most recent creatinine being more than 180 µmol/L (2 mg/dL) prior to transplantation, were not significant risk factors. These data suggest that the critically urgent patients who survived the first year had no increased mortality risks in subsequent years. Other factors that were significant in the 1-year model but not significant in the conditional 3-year model were donor race and blood type compatibility.

Heart Graft Survival

In contrast to the many factors for kidney and liver survival, only 4 donor risk factors were found to affect 1-year heart graft survival(Table 7). The mean donor age for hearts (26 years) was less than that for kidney (32 years) and liver (28 years) donors. For the first 10-year increment in donor age, the odds of 1-year graft failure increased 14%. For the first 20-year increment in donor age, the odds of 1-year graft failure increased 43%. Preservation time of hearts was much more limited and 1 additional hour of graft preservation increased the odds of graft failure by 9%. Furthermore, the odds of graft failure was higher for minority donors and for female donors than for white male donors.

The most significant cardiac recipient risk factor was undergoing a repeat transplantation, which more than tripled the likelihood of graft failure at 1 year after transplantation. The recipient's immediate preoperative status was a strong predictor of short-term outcome: patients who were in an intensive care unit had a 13% increased odds of 1-year graft failure; patients supported by an intra-aortic balloon pump, a ventricular assist device, or a ventilator had respectively an 18%, 33%, and 100% higher odds of 1-year graft failure. The odds of graft failure for black recipients and female recipients was significantly higher than for white males. Congenital heart disease had a 34% increase in the odds of graft failure compared with coronary artery disease and cardiomyopathy. A 10-year increment in recipient age increased the odds of graft failure by 15%, while a 20-year increment in recipient age increased the odds of graft failure by 64%. Patients receiving transplants in 1993 to 1994 had significantly improved graft survival compared with those in 1988 to 1992. Although required support by any of the assist devices (intra-aortic balloon pump, ventricular assist device, ventilator) was associated with poorer short-term survival, it did not have a significant effect if the patient survived the first year after transplantation. The only factors found to affect long-term outcomes were donor age, donor race, and recipient race.

Center-Specific Results

For each organ type and transplant program in the United States, the 1997 center-specific report details the program's 3-month (1-month for thoracic organs), 1-year, 3-year, and conditional 3-year actual graft and patient survival rates, the risk-adjusted expected survival rates, the differences between actual and expected survival rates, and the P values of the differences.5 In addition, the report depicts the frequencies of each transplant program's donor and recipient characteristics (risk factors) and compares them with the frequencies of national donor and recipient risk factors that were used in the model to calculate expected graft and patient survival rates. Each individual program director had the opportunity to review his/her program's results and make comments about the data. The program directors' narratives are published in the 1997 report.5

Categorization of Programs' Actual Graft and Patient Survival as Being Above, Below, or Equal to Expected Rates

For each transplant program, actual 3-month (1-month), 1-year, and conditional 3-year graft and patient survival were determined to be above, below, or equal to its risk-adjusted expected survival. These categories were based on the corresponding P value of the difference between the expected and actual survival rates. The categorization of each organ program's 1-year and conditional 3-year graft survival are depicted in Figure 1. For most programs, the actual survival rate was not significantly different from the expected rate. Programs with actual 1-year graft survival rates significantly less than expected rates ranged from 24.3% of liver programs to 1.7% of heart-lung programs. Programs with actual conditional 3-year graft survival rates significantly less than the expected rates ranged from 10.6% of kidney programs to 3% of heart-lung programs. The percentage of programs with less-than-expected survival rates at conditional 3-years were much smaller than the percentages at 1 year. The larger differences in actual and expected rates, for all organs, were nearly always seen in programs that reported relatively few transplantations. This is best seen for kidney transplantations where the difference between actual and expected survival is plotted against the average number of transplantations per year. Nearly all differences greater than 10 percentage points (either positive or negative) were found among programs performing fewer than 25 kidney transplantations per year. Similar observations were made for liver, heart, and lung programs (the 1997 center-specific report presents similar scatterplots for all organs).5

The 1994 center-specific report identified 21 kidney, 13 liver, and 16 heart transplant programs whose actual 1-year graft survival rates were significantly less than expected.2 The current report shows that in the majority of these programs, 1997 performance was also significantly less than expected. In the 15 kidney transplant programs whose performance was less than expected in both the 1994 and 1997 reports, 8 performed fewer than 40 transplantations (the national average) and 5 performed fewer than 25 per year. In the 13 liver transplant programs whose performance was fewer than expected in both reports, 9 performed less than 26 transplantations per year (the national average) and 6 performed fewer than 10 per year. One of these programs is a pediatric hospital, where it may have been difficult to achieve a threshold volume, and 3 programs are currently inactive. In the 15 heart transplant programs whose performance was less than expected in both reports, 9 performed fewer than 12 transplantations per year (the national average). Three of these programs are pediatric hospitals and 3 additional programs are currently inactive.

For the 38 new programs that started after the cohort in the 1994 report (October 1987-December 1991) (13 kidney, 15 liver, and 10 heart), only 1 of the heart transplant programs had actual results less than expected.

COMMENT

The 1997 Report of Center Specific Graft and Patient Survival Rates is unique in that it includes every orthotopic solid organ allotransplantation (except multivisceral transplantations) performed in the United States from January 1988 through April 1994. This includes all transplantations performed at government hospitals, including the Veterans Affairs hospitals; university hospitals, both public and private; and for-profit and not-for-profit private hospitals. Other large databases have been assembled to assess risk-adjusted surgical results,6,7 including center-specific outcomes,8,9 and, in some instances, to assess both center-specific and surgeon-specific outcomes,1012 but they have evaluated only short-term surgical mortality. The limitations of using hospital mortality as the primary yardstick of quality of care have previously been examined and discussed.13

The 1997 report is also unique because, despite the magnitude of this project (97587 solid organ transplantations) and the fact that reporting data to UNOS by the 742 individual transplant programs is voluntary, 1-year graft survival follow-up exceeded 98% for all organs. Conditional 3-year follow-up for both graft and patient survival exceeded 91% for all organs. All data compiled by UNOS were validated by the individual contributing programs. Additionally, UNOS auditors perform on-site reviews of transplant centers and compare the validated data with raw data from patients' records for accuracy.4 The auditors detected an overall error rate of 0.5%, which speaks to the dedication of the personnel of the transplant programs to maintaining a high-quality database in the UNOS Scientific Registry.

The survival analyses performed for each organ in the 1997 report demonstrated improved graft and patient survival rates between era 1 and era 2 for all cadaveric organs except the heart (Table 4). While factors responsible for improved survival were not formally examined, there was a trend toward more factors that would negatively affect survival, such as increased donor age for all organs14 and increased recipient age for all organs.4,5 Other negative factors included increased numbers of black donors, donors' dying of cerebrovascular causes, and heart recipients receiving life support.4 Positive factors for improved survival included a decreased number of sensitized kidney recipients, an increased number zero mismatched kidneys, and a decreased number of liver recipients receiving life support.4 Variation in drug protocols may have affected both center results as well as national results in era 1 and era 2. While most patients during the overall period received cyclosporine-based immunosuppressive protocols,15 tacrolimus16 was introduced into some centers' drug protocols, and there was an increased use of polyclonal and monoclonal antibody induction therapy only during the latter portion of this study.15 The 1997 report does not investigate drug protocols used by individual transplant programs because of the lack of uniformity and frequent changes in individual protocols. The 1997 report is the first center-specific report that used national data to evaluate separately short-term and long-term risk factors. In addition to the 1-year survival rate, the 1997 report shows the conditional 3-year survival rate (ie, the 3-year survival rate conditional on surviving at least 1 year after transplantation) both nationally and by individual transplant center. This conditional analysis permitted an assessment of characteristics independent of those limited to the first year, such as the risk of surgical procedures, delayed graft function, and early acute rejection events. The fact that the conditional 3-year graft survival rates were generally higher than the 1-year graft survival rates indicated that the first year after transplantation was more critical to the ultimate success of the graft than the subsequent 2 years. Furthermore, if a transplant recipient can survive 1 year, the long-term survival outlook is generally optimistic (Table 3). The results from the 1997 report also indicated that center effects, in terms of the significant difference between the actual and expected survival rates, were mostly manifested within the first year after transplantation and were much less influential on long-term survival outcomes. This is supported by conditional 3-year survival rates of more than 90% for kidney, liver, pancreas, and heart transplantations (Table 3). This also was evident from the percentages of programs with actual survival rates less than the expected survival rates. For kidney, liver, pancreas, and heart, these percentages were smaller in the 3-year conditional model than they were in the 1-year model(Figure 1).

Opelz et al17 first described the center effect in 1975 when they analyzed data from 95 kidney transplant programs. They defined outliers as those programs whose survival rates were outside 2 SDs from the mean. Terasaki's University of California, Los Angeles (UCLA) group,1821 using different definitions, further described the center effect on multiple occasions. Mickey18 compared kidney transplant centers' actual survival rates with their expected rates and showed that about one third of the observed variations were due to the center effect. The 1997 report and the UCLA studies1721 are different in that the 1997 report includes different covariates and reports both an actual and a risk-adjusted expected rate for each individual program. The 1997 report and the UCLA studies are similar in that both note a more prominent center effect in the short term. In addition, both suggest that survival models consisting of donor and patient-mix variables alone can explain only a limited amount of variation in center-specific survival rates. In the 1997 report, a modified R2 is used to measure the degree of program-to-program variation in actual survival rates that can be accounted for by donor and recipient risk factors.5 The R2 for 1-year graft survival models was 38% for liver, 36% for heart, 36% for lung, 34% for pancreas, 22% for kidney, and 18% for heart-lung. These results suggest that there are factors in addition to those included in the survival models that could improve the prediction of survival rates. Center effects may be most important among them.

Although a large number of donor and recipient risk factors were evaluated for each organ, cardiac and pulmonary risk factors potentially important to surgical mortality6,22 were not evaluated for abdominal organ transplants (ie, kidney, liver, pancreas), because the data were not collected by UNOS. It has been shown previously that preoperative cardiac risk evaluation and risk stratification for kidney transplant recipients significantly predicts postoperative cardiac mortality.23 Factors such as socioeconomic status and insurance coverage, previously shown to influence graft survival,24 also were not considered in the 1997 report. Additionally, as mentioned before, a major factor not considered in the 1997 report was the individual transplant center's immunosuppressive regimen. The analysis by Katznelson and Cecka15 of UNOS data on 88883 kidney transplantations showed no advantage of one cyclosporine-based drug protocol over another, including incorporation of induction therapy with antilymphocyte globulin or muromonab-CD3 (Orthoclone OKT-3), although Cecka et al25 earlier reported antilymphocyte globulin or muromonab-CD3 induction was beneficial especially when there was delayed function of the transplant. A multivariate study by Shield et al26 reports a beneficial effect on both anitlymphocyte globulin and muromonab-CD3 induction therapy under defined dosage regimens. At any rate, since the primary objective of the report is to assess the quality of each individual transplant program, ie, the center effect, including the drug protocol in the survival models, which might have confounded the measurement of the center effect because the use of drugs reflects the practice of a program.

Transplant volume differed greatly among programs. The majority of transplantations were performed by a small number of programs. For example, nearly 75% of all liver transplants were performed by 27% of the 103 liver programs, and approximately 20% of the 249 kidney programs performed 50% of all kidney transplantations during the study period.

It appears that the number of transplantations performed at an individual program is an important characteristic to consider when evaluating survival rates. In general, the more transplantations a program has performed, the more accurately the survival rates reflect the quality of the program. For kidney transplantations, most programs whose actual rates deviated more than 10 percentage points from expected rates were among those performing fewer than 25 transplantations per year. This finding is similar to an earlier UNOS study, which demonstrated that the risk of mortality at early and intermediate time points is higher in low-volume cardiac transplant centers.27 A relationship between procedure volume and surgical mortality has previously been shown for coronary artery bypass procedures, abdominal aortic aneurysm resections, and other general surgical procedures.28

In evaluating poor performance transplant programs, the 1997 report does not specifically evaluate the programs for the 27-month period beginning at the end of the 1997 report cohort (April 1994) to the present time. Therefore, some of these programs may have experienced a substantial improvement in performance that was not determined in this study. Nevertheless, any outlier program identified in the 1997 report will be reviewed further by the UNOS Membership and Professional Standards Committee using a more recent cohort. If a program's performance continues to fall below the objective criteria developed by the committee, it will be site visited by committee designees. The objective criteria used by the committee incorporate both statistical significance and clinical significance by using a performance threshold line. The line is determined by the P value of the deviation (D) of the actual survival rate from its expected survival rate (D=−25×P−10). For P values that are highly significant (P≈0), the performance threshold is approximately a 10% decrease from the expected rate. For P values that are highly nonsignificant (P≈1) but considered to be clinically significant, the performance threshold is about a 35% decrease from the expected rate.3 Based on the results of the 1994 report and its subsequent analysis, 2 kidney, 7 liver, and 8 heart programs were site visited. During these visits, several organizational practices and procedures were evaluated, including program protocols, the distribution of clinical responsibilities, and transplant clinic organization and staffing. The visits were well received and productive.3 A similar approach of site visiting certain hospitals has been described for coronary artery bypass surgery and was felt to have resulted in reduced mortality rates.12

References
1.
Edwards E. 1991 Center Specific Graft and Patient Survival Rates . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1992.
2.
Edwards E, Bennett L, Guo T.  et al.  1994 Center Specific Graft and Patient Survival Rates . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1994.
3.
Burdick J, Norman DJ, Hunsicker L.  et al.  Identification of poorly performing transplant centers using the UNOS center-specific data.  Transplant Proc.1997;29:1495.
4.
Smith CM, Ellison MD, White R.  et al.  1996 Annual Report of the US Scientific Registry for Transplant Recipients and the Organ Procurement and Transplantation Network . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1996.
5.
Lin H-M, McBride MA, Davies D.  et al.  1997 Center Specific Graft and Patient Survival Rates . Washington, DC: US Dept of Health and Human Services; Richmond, Va: United Network for Organ Sharing; 1997.
6.
Copeland GP, Jones D, Walters M. POSSUM: a scoring system for surgical audit.  Br J Surg.1991;78:355-360.
7.
Grover FL, Hammermeister KE, Burchfiel C.  et al.  Initial report for the Veterans Administration preoperative risk assessment study for cardiac surgery.  Ann Thorac Surg.1990;50:12-28.
8.
Edwards FH, Clark RE, Schwartz M. Coronary artery bypass grafting.  Ann Thorac Surg.1994;57:12-19.
9.
Khuri SF, Dailey J, Henderson W.  et al.  Risk adjustment of the postoperative mortality rate for the comparative assessment of the quality of surgical care: results of the National Veterans Affairs Surgical Risk Study.  J Am Coll Surg.1997;185:315-327.
10.
Hannan EL, Kilburn H, Racz M.  et al.  Improving the outcomes of coronary artery bypass surgery in New York State.  JAMA.1994;271:761-766.
11.
Hannan EL, Siu AL, Kumar D.  et al.  The decline in coronary artery bypass graft surgery mortality in New York State: the role of surgeon volume.  JAMA.1995;273:209-213.
12.
O'Connor GT, Plume SK, Olmstead EM.  et al.  A regional intervention to improve the hospital mortality associated with coronary artery bypass surgery.  JAMA.1996;275:841-846.
13.
Wennberg JE, Freeman JL, Shelton RM.  et al.  Hospital use and mortality among Medicare beneficiaries in Boston and New Haven.  N Engl J Med.1989;321:1168-1173.
14.
Kauffman HM, Bennett LE, McBride MA, Ellison MD. The expanded donor.  Transplant Rev.1997;11:165-190.
15.
Katznelson S, Cecka JM. Immunosuppressive regimens and their effects on renal allograft outcome. In: Cecka JM, Terasaki PI, eds. Clinical Kidney Transplants, 1996 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1997:361.
16.
Klintmalm G. A review of FK506: a new immunosuppressant agent for the prevention and rescue of graft rejection.  Transplant Rev.1994;8:53-63.
17.
Opelz G, Mickey MR, Terasaki PI. Comparison of kidney transplant survival among transplant centers.  Transplantation.1975;19:226-229.
18.
Mickey MR. Center variability. In: Terasaki PI, ed. Clinical Kidney Transplants, 1989 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1989:435-446.
19.
Cicciarelli J. Transplant center and kidney graft survival. In: Terasaki PI, ed. Clinical Kidney Transplants, 1985 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1986:93.
20.
Cho YW, Cecka JM. Center effect in the UNOS renal transplant registry. In: Terasaki PI, ed. Clinical Kidney Transplants, 1992 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1993:333.
21.
Cho YW, Cecka JM. Organ procurement organization and transplant center effects on cadaver renal outcomes. In: Terasaki PI, ed. Clinical Kidney Transplants, 1996 . Los Angeles: University of California, Los Angeles, Tissue Typing Laboratory; 1997:427.
22.
Khuri SF, Daley J, Henderson W.  et al.  The national Veterans Administration Surgical Risk Study.  J Am Coll Surg.1995;180:519-531.
23.
Le A, Wilson R, Douek K. Prospective risk stratification in renal transplant candidates for cardiac death.  Am J Kidney Dis.1994;24:65-71.
24.
Butkus DE, Meydrech EF, Raju SS. Racial differences in the survival of cadaveric renal allografts: overriding effects of HLA matching and socioeconomic factors.  N Engl J Med.1992;327:840-845.
25.
Cecka JM, Gjertson D, Terasaki PI. Do prophylactic antilymphocyte globulins (ALG and OKT3) improve renal transplant survival in recipient and donor high-risk groups?  Transplant Proc.1993;25:548-549.
26.
Shield CF, Edwards EB, Davies DB, Daily OP. Antilymphocyte induction therapy in cadaver renal transplantation.  Transplantation.1997;63:1257-1263.
27.
Hosenpud JD, Breen TJ, Edwards EB.  et al.  The effect of transplant center volume on cardiac transplant outcome.  JAMA.1994;271:1844-1849.
28.
Hannan EL, O'Donnell JF, Kilburn H.  et al.  Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals.  JAMA.1989;262:503-510.
×