Compared with deceased donor kidney transplantation, live donor kidney transplantation (LDKTx) offers longer patient and allograft survival and greater opportunities for preemptive transplantation (1–3). In the United States, the volume of live kidney donation more than doubled from 1994 to 2006, but has since declined (4). During this growth period, multiple innovations took place, including laparoscopic nephrectomy, greater use of “unrelated” donors (who are not first-degree family members or spouses of the recipient), and the development of methods to overcome biological incompatibility between donor-recipient pairs (5–10). Assessment of variation in the magnitude of LDKTx between centers could be an important step toward identifying center practices that facilitate live kidney donation.
Centers vary widely in the proportion of kidney transplants performed using live donors, although a valid comparison of LDKTx rates would require adjustment for the demographics of the center's population (11). For instance, older age and black race are associated with a lower likelihood of LDKTx (11–19). Although centers cannot easily change the populations they serve, a center may facilitate LDKTx through innovative programs, such as recipient desensitization or living donor exchange (5, 6, 20). These approaches are hypothesized to increase LDKTx overall, but their expense and time demands may impede other important center work such as new donor evaluations. The effects of other practices (such as using unrelated donors) and other center attributes (such as volume) on LDKTx have not been fully explored.
The study had three aims. The first aim was to develop a standardized live donor transplantation ratio (SLDTR) that would assess center differences in LDKTx rates after adjustment for center population. The second and third aims were to identify center attributes associated with greater odds that an individual candidate would undergo LDKTx, and center attributes associated with greater odds that a center was consistently in the upper three quartiles of SLDTR.
The primary cohort consisted of 148,168 individuals at 194 centers. A total of 34,593 (23.3%) individuals underwent LDKTx within 1.5 years; 39.3% of these recipients were never wait listed. Among transplant candidates who were wait listed and underwent LDKTx, the distribution of wait-list time was right-skewed, with a median of 162 days (∼5.4 months).
Table 1 presents characteristics of individuals who did and did not meet the outcome of LDKTx within 1.5 years. In unadjusted analyses, older age, black race, Hispanic ethnicity, diabetes, elevated panel reactive antibody (PRA), type O blood, lower education, and nonprivate health insurance were associated with a lower likelihood of LDKTx (P<0.05 for all variables). These characteristics were used to generate strata for indirect standardization, to calculate the expected number of live donor transplants at each center and to calculate each center's SLDTR during each year. Because previous studies have suggested an association between LDKTx and sex, we also used sex for the indirect standardization (21).
Transplant Center Characteristics
The mean annual number of live donor kidney transplants at a center was 25.5 (range 0.7–153.6; Table 2). The mean proportion of unrelated donors/all live donors was 8.6%. Eighty-four (43.3%) centers performed any transplants between incompatible donor-recipient pairs.
Multivariable Analysis of Individual and Center Characteristics Associated With the Outcome of Undergoing LDKTx
All individual characteristics associated with a lower likelihood of LDKTx in univariate analysis remained significant in multivariable logistic regression, except for Hispanic ethnicity (Table 3). Notably, male sex was associated with a lower odd ratios of LDKTx. A significant interaction between black race and male sex (odds ratio [OR] 0.74; P<0.01) indicated that black men faced a particularly high barrier to finding a live donor.
Higher center use of unrelated donors (OR 1.31 for highest tertile vs. lowest tertile of unrelated donors; P=0.02) and having a program to overcome donor-recipient incompatibility (OR 1.33; P=0.01) were associated with greater individual access to LDKTx.
Multivariable Analysis of Center Characteristics and Center Ranking by Magnitude of LDKTx
The SLDTR had a mean of 1.1 (SD 0.5) and median of 1.0. During the 7 years of the study, the mean SLDTR remained close to 1, but the upper bound of the SLDTR increased from 2.5 to 3.6.
One hundred forty-six centers (75.3%) were in the upper three SLDTR quartiles for at least 4 years. Centers in the highest three quartiles in any year were likely to be in the highest three quartiles during another year. For example, compared with centers in the lowest quartile in 2003, centers in the highest three quartiles in 2003 had an OR of 90.0 (P<0.01) for being in the highest three quartiles again during 2004, 2005, or both.
In multivariable logistic regression, centers with a program to overcome donor-recipient incompatibility (OR 4.79; P<0.01), with greater use of unrelated donors (OR 8.30 per higher tertile of unrelated donor use; P<0.01), and with greater use of laparoscopic nephrectomy (OR 2.53 per higher tertile of laparoscopic nephrectomy; P=0.02) were more likely to be in the upper three SLDTR quartiles for at least 4 years. Table 4 shows these results.
Potential for Expansion of LDKTx
We estimated that an additional 766 transplants (mean 15.6 per low center) would take place if centers in the lowest quartile during 2005 performed the expected number of transplants.
In a linear regression model, we examined the secondary outcome of a center performing a higher proportion of (live donor transplants)/(all kidney transplants). Greater waiting time (β=0.06 per tertile of higher waiting time; P<0.01), having a donor-recipient incompatibility program (β=0.06; P<0.01), using a higher proportion of unrelated donors (β=0.06 per tertile of unrelated donor use; P<0.01), and greater use of laparoscopic nephrectomy (β=0.03 per tertile of laparoscopic nephrectomy; P<0.01) were all associated with this outcome, whereas higher center volume was inversely associated with this outcome (β=−0.01; P=0.03).
Our analysis revealed wide variation in the magnitude of LDKTx across centers after adjustment for their patient populations. Center use of unrelated live donors and the presence of a biological incompatibility program were associated with greater odds that an individual candidate would undergo LDKTx. Centers with greater use of unrelated donors, with a biological incompatibility program, and with greater use of laparoscopic nephrectomy were less likely to lag behind others in the magnitude of LDKTx. These findings may help guide centers toward practices that expand access to LDKTx.
Our study confirmed substantial disparities in access to LDKTx associated with black race and older age. The findings related to race may be due to the higher prevalence of kidney disease and poverty in this population, which could limit the number of medically suitable donors who can manage the financial burden of donation (14, 22, 23). We hypothesize that older candidates feel protective of potential donors and discourage them. Potential donors may also sense that transplantation confers less benefit to older recipients (12, 18).
Higher education and private insurance were associated with greater individual access to LDKTx. Education and health literacy may be important to recognizing the benefits of LDKTx. This finding underscores the need for transplant programs to provide information that is easily understood by individuals with limited literacy (24, 25). It is also possible that private insurance is simply a marker for greater wealth, social status and support, and that patients with these means can attract live donors.
The SLDTR metric showed evidence of stability. We focused particular attention on the contrast between centers in the lowest quartile—where transplant candidates have minimal access to LDKTx—versus centers consistently in the upper three quartiles of SLDTR. Centers in the highest three quartiles were highly likely (OR 90.0) to remain in those quartiles the next year. This finding suggests that a center's ranking was not usually due to random variation, but rather to modifiable processes of care.
Center practices may increase the magnitude of LDKTx. Our analysis confirms the importance of donor-exchange and blood group incompatible transplant in expanding overall use of LDKTx (5, 26). However, these programs are resource intensive. Desensitization of blood group incompatible pairs (and of recipients with antibodies against donor human leukocyte antigens) also may heighten risks of infection and malignancy. On the other hand, the acceptance of unrelated donors is neither more risky nor ethically problematic than accepting related donors. Encouraging transplant centers to expand the use of unrelated donors could be an underrecognized way of maximizing LDKTx. We acknowledge that unmeasured center attributes are also likely to drive variation in LDKTx rates. Some centers may have staff with greater personal interest in working with live donors; other centers may devote greater resources to educational initiatives to raising awareness about the benefits of LDKTx. Interestingly, a metric of market competition (defined by donor service area) was unrelated to the SLDTR. It is possible that many centers actually compete beyond region and that the proximity of local centers has little effect on their practices with respect to LDKTx. It is also plausible that the financial benefits that centers gain from increasing live donor transplant drives their practice more than market competition.
Our study has limitations. First, the selection of live donors must respect their health and interests. Our group and others have voiced concern about accepting “medically complex live donors”—donors with risk factors for kidney disease such as obesity (27–30). One troubling way that centers may increase LDKTx is through relaxing donor medical criteria. Because of missing relevant data, we were not able to examine how the use of medically complex donors may explain SLDTR variation.
We also used multiple imputation to account for missing data about PRA and the date of transplant readiness among live donors who were never wait listed. Imputation may create bias, although sensitivity analyses suggest that imputation did not affect our main results.
We acknowledge the methodological limitation of excluding early recipients of deceased donor transplants from the SLDTR calculation. We developed this approach because the receipt of a deceased donor kidney within only 1.5 years of entering the wait list would be considered a favorable outcome for almost any patient. Centers should not necessarily be “penalized” for not facilitating LDKTx for individuals who rapidly received deceased donor allografts. Removing these early deceased donor recipients allowed the analysis to focus on candidates with extended periods on the wait list, during which patients often suffer deteriorating health. Also, our approach effectively allowed us to account for differences in waiting time and access to deceased donor allografts between regions. As Segev et al. (31) have shown, patients and centers in regions with a short waiting time may be less likely to pursue LDKTx. In the future, if further development of methods to compare LDKTx rates became a priority for the United Network for Organ Sharing (UNOS), centers might be required to report the date of transplant readiness for all candidates; the SLDTR could then be calculated using a time-to-event approach in which deceased donor kidney recipients were censored at transplantation.
Our analysis also does not provide insight into events before “transplant readiness.” Some centers may be efficient at completing the donor workup, but our dataset did not allow us to study the effect of center practices on candidate outcomes before wait listing or actual transplantation. Finally, our study did not measure the impact of educational programs about LDKTx for referring nephrologists, primary care doctors, or the general public; these efforts may also increase donation rates.
Variation in LDKTx across centers is only partly explained by differences in center populations and practices. Comparing rates of LDKTx provides an opportunity to identify those centers that do not maximize opportunities for candidates to receive a live donor transplant. In particular, developing programs such as live donor exchange that overcome biological incompatibility, and increasing the acceptance of unrelated donors, may enable expansion of LDKTx.
MATERIALS AND METHODS
Generation of Cohorts
We used UNOS data to assemble seven cohorts of adult (≥18 years) candidates considered eligible for LDKTx corresponding to the years 1999 to 2005 (Fig. 1). We defined a transplant candidate as any adult added to the kidney transplant wait list during that year. Individuals who underwent LDKTx but were never added to the list were also considered candidates and were assigned a date of transplant readiness of 192 days before transplantation (the median days for recipients who were wait listed).
Measurement of LDKTx
To compare the LDKTx rate between centers, we analyzed live donor transplants within 1.5 years of the date of transplant readiness. This time period was selected because, among individuals who were wait listed and underwent LDKTx, 85% of these transplants took place within 1.5 years.
We excluded patients at centers that performed, on average, less than 10 kidney transplants per year and that did not add anyone to the wait list during every year of the study period. To account for differences in severity of illness between center populations, we excluded patients who died or were removed from the wait list within 6 months. To account for regional variation in the availability of deceased donor allografts, we removed any patient who underwent deceased donor kidney transplantation within 1.5 years of wait listing. Removing deceased donor transplant recipients was consistent with our primary focus on patients who languish on the wait list without access to transplantation.
Transplant Candidate Characteristics
On the basis of previous studies and our clinical experience, we examined associations between the outcome of an individual undergoing LDKTx and these characteristics: age, black race, Hispanic ethnicity, sex, diabetes, elevated PRA (a binary variable defined by UNOS convention as ≥80% or not), blood group (a binary variable defined as type O or not), education (defined as no college education, some college education, and completing college), and health insurance (private, Medicare, Medicaid, or other) (11, 13, 18, 21). Age was examined as an ordinal variable (<40, 40–60, and >60 years) because the association with LDKTx was not linear.
These center characteristics were evaluated: use of “unrelated” donors (the mean proportion of unrelated donors/all live donors); use of laparoscopic nephrectomy (the mean proportion of laparoscopic nephrectomies/all live donor nephrectomies); performing any LDKTx across incompatible blood groups; performing any donor exchanges; center volume of kidney transplants; and market concentration. Market concentration of transplant centers was assessed using the Herfindahl-Hirschman Index (32) calculated as the sum of the squared market shares of centers in a donor service area (32, 33). We converted the Herfindahl-Hirschman Index into a binary variable using a traditional cut-point from economics (≤0.10 is high competition, >0.10 is low competition) (34, 35).
Because the variables of laparoscopic nephrectomy use, center volume, and proportion of nonfamily donors were skewed, we converted these variables to ordinal tertiles. Center practices of ABO-incompatible transplant and donor exchange were correlated (P<0.01), so we categorized centers performing either practice as having a biological incompatibility program.
Multivariable Analysis for the Outcome of an Individual Undergoing LDKTx (Aim 1)
We fit a multivariable logistic regression model for the outcome of an individual undergoing LDKTx and entered independent center and patient characteristics. Given a study of access to kidney transplantation that showed an age-gender interaction, we explored for interactions between age and gender, age and race, and gender and race (12).
Generation of an Adjusted Metric of Center-Specific Rate of LDKTx
Using the individual characteristics that were significantly associated with LDKTx in univariate analysis, we used indirect standardization to determine strata within the population of transplant candidates nationally (36). Indirect standardization has been used to calculate expected rates of medical events such as deaths among dialysis patients (37, 38). We applied stratum-specific rates of LDKTx to each center's population to derive the expected number of live donor transplants per year. We then calculated the ratio of observed transplants to expected transplants at that center (the SLDTR).
Multivariable analysis of center characteristics and the outcome of center ranking by magnitude of LDKTx (Aim 2): We ranked centers by SLDTR and divided them into quartiles each year. Centers were then categorized according to the binary outcome of whether the center had a consistently low magnitude of LDKTx, defined as being in the lowest SLDTR quartile for the majority of the study period (i.e., at least 4 of 7 years). We focused on identifying characteristics of centers that performed fewer live donor transplants, because these centers have the greatest opportunity to expand the practice. Also, centers repeatedly in the lowest quartile would be less likely to have a low SLDTR due to temporary circumstances (such as departure of a transplant surgeon).
To maintain consistency in the direction of the ORs between our first and second analyses, the outcome for the second logistic regression model was being a center in the upper three quartiles of SLDTR for at least 4 years. For example, if a center's use of unrelated donors increased individual access to LDKTx and raised a center's SLDTR ranking, this practice would be associated with an OR more than 1 in both analyses.
A minority of patients had missing data on education and income. Additionally, some wait-listed patients and recipients of live donor transplants who were not wait listed lacked data on peak PRA (n=20633, or 13.9% of the cohort). For our multivariable analyses, we performed multiple imputation to account for missing data.
We fit a multivariable linear regression model for the outcome of the center-specific proportion of live donor kidney transplants (live donor transplants/all transplants). We evaluated the same independent variables used in the primary analysis and median waiting time in the center's donor service area.
We also calculated an SLDTR using a “measurement window” of 2.5 years within which LDKTx took place and repeated multivariable regression analyses. Additionally, we performed two sensitivity analyses in which the imputed days between transplant readiness and transplant was 84 days (the 25th percentile for number of days between wait listing and LDKTx for recipients who were wait listed) and 274 days (the 75th percentile).
Analyses were performed using SAS (version 8, SAS Institute Inc., Cary, NC) and Stata (version 10.0, Stata Corporation, College Station, TX). For unadjusted comparisons of continuous variables, the t test or the rank-sum test were used, as appropriate. The chi-square test was used to compare categorical variables. The primary multivariable analyses were performed using logistic regression; the Hosmer-Lemeshow hypothesis of good fit was not rejected (P>0.05).
1. Davis CL, Delmonico FL. Living-donor kidney transplantation: A review of the current practices for the live donor. J Am Soc Nephrol
2005; 16: 2098.
2. Kasiske BL, Snyder JJ, Matas AJ, et al. Preemptive kidney transplantation: The advantage and the advantaged. J Am Soc Nephrol
2002; 13: 1358.
3. Organ Procurement and Transplantation Network. Kidney Kaplan-Meier Graft Survival Rates For Transplants Performed: 1997–2004. Available at: http://optn.transplant.hrsa.gov/latestData/rptData.asp
. Accessed October 5, 2009.
4. Organ Procurement and Transplantation Network. Donors Recovered in the U.S. by Donor Type. Available at: http://optn.transplant.hrsa.gov/latestData/rptData.asp
. Accessed: October 5, 2009.
5. Segev DL, Gentry SE, Warren DS, et al. Kidney paired donation and optimizing the use of live donor organs. JAMA
2005; 293: 1883.
6. Takahashi K, Saito K, Takahara S, et al. Excellent long-term outcome of ABO-incompatible living donor kidney transplantation in Japan. Am J Transplant
2004; 4: 1089.
7. USRDS. 2008 Annual Report, Figure 7.2, Vol. 2: Counts of transplants from living donors, by donor relation, National Institutes of Health, National Institutes of Diabetes and Digestive and Kidney Diseases, Bethesda, MD.
8. Ratner LE, Ciseck LJ, Moore RG, et al. Laparoscopic live donor nephrectomy. Transplantation
1995; 60: 1047.
9. Montgomery RA, Zachary AA, Racusen LC, et al. Plasmapheresis and intravenous immune globulin provides effective rescue therapy for refractory humoral rejection and allows kidneys to be successfully transplanted into cross-match-positive recipients. Transplantation
2000; 70: 887.
10. Tanabe K, Takahashi K, Sonda K, et al. Long-term results of ABO-incompatible living kidney transplantation: A single-center experience. Transplantation
1998; 65: 224.
11. Axelrod DA, McCullough KP, Brewer ED, et al. Kidney and pancreas transplantation in the United States, 1999–2008: The changing face of living donation. Am J Transplant
2010; 10(4 pt 2): 987.
12. Segev DL, Kucirka LM, Oberai PC, et al. Age and comorbidities are effect modifiers of gender disparities in renal transplantation. J Am Soc Nephrol
2009; 20: 621.
13. Epstein AM, Ayanian JZ, Keogh JH, et al. Racial disparities in access to renal transplantation—Clinically appropriate or due to underuse or overuse? N Engl J Med
2000; 343: 1537.
14. Lunsford SL, Simpson KS, Chavin KD, et al. Racial disparities in living kidney donation: Is there a lack of willing donors or an excess of medically unsuitable candidates? Transplantation
2006; 82: 876.
15. Jindal RM, Ryan JJ, Sajjad I, et al. Kidney transplantation and gender disparity. Am J Nephrol
2005; 25: 474.
16. Young CJ, Gaston RS. Renal transplantation in black Americans. N Engl J Med
2000; 343: 1545.
17. Boulware LE, Ratner LE, Sosa JA, et al. The general public's concerns about clinical risk in live kidney donation. Am J Transplant
2002; 2: 186.
18. Gore JL, Danovitch GM, Litwin MS, et al. Disparities in the utilization of live donor renal transplantation. Am J Transplant
2009; 9: 1124.
19. Weng FL, Reese PP, Mulgaonkar S, et al. Barriers to living donor kidney transplantation among black or older transplant candidates. Clin J Am Soc Nephrol
2010; 5: 2338.
20. Ross LF, Rubin DT, Siegler M, et al. Ethics of a paired-kidney-exchange program. N Engl J Med
1997; 336: 1752.
21. Bloembergen WE, Port FK, Mauger EA, et al. Gender discrepancies in living related renal transplant donors and recipients. J Am Soc Nephrol
1996; 7: 1139.
22. Reese P, Oyedeji A, Grossman R. Race and living kidney donors. Transplantation
2007; 83: 1139; author reply 1139.
23. Lentine KL, Schnitzler MA, Xiao H, et al. Racial variation in medical outcomes among living kidney donors. N Engl J Med
2010; 363: 724.
24. Gordon EJ, Wolf MS. Health literacy skills of kidney transplant recipients. Prog Transplant
2009; 19: 25.
25. Gordon EJ, Caicedo JC, Ladner DP, et al. Transplant center provision of education and culturally and linguistically competent care: A national study. Am J Transplant
2010; 10: 2701.
26. Vo AA, Lukovsky M, Toyoda M, et al. Rituximab and intravenous immune globulin for desensitization during renal transplantation. N Engl J Med
2008; 359: 242.
27. Reese PP, Caplan AL, Kesselheim AS, et al. Creating a medical, ethical, and legal framework for complex living kidney donors. Clin J Am Soc Nephrol
2006; 1: 1148.
28. Reese PP, Feldman HI, McBride MA, et al. Substantial variation in the acceptance of medically complex live kidney donors across US renal transplant centers. Am J Transplant
2008; 8: 2062.
29. Young A, Storsley L, Garg AX, et al. Health outcomes for living kidney donors with isolated medical abnormalities: A systematic review. Am J Transplant
2008; 8: 1878.
30. Steiner RW. Risk appreciation for living kidney donors: Another new subspecialty? Am J Transplant
2004; 4: 694.
31. Segev DL, Gentry SE, Montgomery RA. Association between waiting times for kidney transplantation and rates of live donation. Am J Transplant
2007; 7: 2406.
32. Scanlon DP, Hollenbeak CS, Lee W, et al. Does competition for transplantable hearts encourage ‘gaming’ of the waiting list? Health Aff (Millwood
) 2004; 23: 191.
33. Hirschman A. The paternity of an index. Am Econom Rev
1964; 54: 761.
34. United States Department of Justice. The Herfindahl-Hirschman Index. Available at: http://www.usdoj.gov/atr/public/testimony/hhi.htm
. Accessed October 16, 2009.
35. Herfindahl O. Concentration in the US steel industry. Economics. New York, Columbia 1950.
36. Rothman KJ, Greenland S. Modern epidemiology [ed. 2]. Philadelphia, PA, Lippincott-Raven 1998.
37. Wolfe RA. The standardized mortality ratio revisited: Improvements, innovations, and limitations. Am J Kidney Dis
1994; 24: 290.
38. Wright J, Dugdale B, Hammond I, et al. Learning from death: A hospital mortality reduction programme. J R Soc Med
2006; 99: 303.