Patient and Kidney Allograft Survival in Recipients With End-Stage Renal Disease From Amyloidosis : Transplantation

Journal Logo

Original Clinical Science—General

Patient and Kidney Allograft Survival in Recipients With End-Stage Renal Disease From Amyloidosis

Sawinski, Deirdre MD1; Lim, Mary Ann MD1; Cohen, Jordana B. MD, MSCE1,2; Locke, Jayme E. MD, MPH3; Weiss, Brendan MD4,5; Hogan, Jonathan J. MD1,4; Dember, Laura M. MD1,2,4

Author Information
Transplantation 102(2):p 300-309, February 2018. | DOI: 10.1097/TP.0000000000001930

Amyloidosis describes a group of disorders in which extracellular deposition of protein as insoluble fibrils results in tissue and organ dysfunction. There are over 30 types of amyloidosis defined by the precursor protein that forms amyloid fibrils.1 Immunoglobulin light chain (AL) amyloidosis is the most common type of amyloidosis affecting the kidney and results from amyloidogenic monoclonal immunoglobulin light chains that are produced, most commonly, by clonal plasma cells.2 AL amyloidosis is rare and is estimated to affect 5 to 12 people/million per year.2,3 Although declining in developed countries, secondary (AA) amyloidosis is an important subtype in patients with autoinflammatory, autoimmune, and chronic infectious diseases.2,4 Serum amyloid A, an acute phase reactant produced by the liver, is the amyloidogenic precursor protein in AA disease.2,4 Other types of amyloidosis affecting the kidney are leukocyte cell-derived chemotaxin 2, fibrinogen A alpha, lysozyme, and, rarely, transthyretin.5

In the absence of therapy, amyloidosis-associated kidney disease frequently progresses to end-stage renal disease (ESRD), with the most rapid rate of deterioration seen in those with AL amyloidosis.6-8 The therapy for amyloidosis is generally geared toward eliminating production of the precursor protein. In AL amyloidosis, the treatment target is the underlying plasma cell clone, and in AA amyloidosis, the target is the underlying inflammation.9-17 In the last 2 decades, the introduction of autologous stem cell transplantation and the development of several new antiplasma cell agents have markedly improved the survival of AL amyloidosis patients.10-15 Similarly, with better control of underlying infectious and inflammatory disorders, the median survival rate for patients with AA amyloidosis is now greater than 10 years.16,17

Despite the improved survival of patients with amyloidosis, data on outcomes for kidney transplantation in patients with amyloidosis-associated ESRD are limited and are comprised mainly of single center reports from an era when treatment was much less effective. Most of the studies suggest that outcomes after kidney transplantation are worse for patients with amyloidosis compared with the general population.18-24 Kidney transplantation for patients with amyloidosis-associated ESRD remains controversial because of: (1) concern for poor allograft function due to recurrent amyloid deposition, (2) anticipated poor allograft and patient outcomes resulting from extrarenal amyloid deposits; and (3) for AL amyloidosis, uncertainty regarding the effects of immunosuppression on underlying clonal cells and potential challenges in managing hematologic relapse. Because of the limited information about kidney transplant outcomes for patients with amyloidosis, we used national transplant registry data to assess patient and kidney allograft survival in patients with ESRD attributed to amyloidosis.


Study Design

We performed a retrospective cohort analysis using national registry data collected by the United Network for Organ Sharing (UNOS), using propensity score matching and Cox proportional hazards modeling. This study is based on Organ Procurement and Transplantation Network (OPTN) data as of March 4, 2016. The database includes information on all transplant recipients and donors in the United States, submitted by the members of the OPTN. The Health Resources and Services Administration, US Department of Health and Human Services provides oversight to the activities of the OPTN contractor. The study met eligibility criteria for institutional review board exemption authorized by 45 CFR §46.101, category 4, as confirmed by the institutional review board at the University of Pennsylvania.


The cohort included all patients transplanted between October 1, 1987, and December 31, 2015. Patient follow-up was through March 4, 2016. The cohort was restricted to adult transplant recipients (age ≥ 18 years) receiving their first kidney transplant; recipients of multiorgan transplants were excluded due to differences in donor and recipient selection practices for these specific populations.

Exposures and Outcome Measures

The primary exposure was defined as ESRD attributed to amyloidosis. The variables “diag_ki” and “diag_ostxt_ki” in the OPTN kidney and pancreas transplant primary data set (KIDPAN_DATA) were used to identify patients with amyloidosis; these variables do not distinguish between amyloidosis subtypes. The primary outcomes were all-cause mortality and all-cause allograft failure; they were determined based on mortality and allograft loss data provided in the OPTN data set.


Covariates that are known risk factors for mortality or allograft loss based on clinical judgment and previously published literature were selected a priori.36-39 Donor-associated covariates included donor type (living vs deceased), Kidney Donor Profile Index (KDPI) and cold ischemia time. Recipient-associated covariates included age, sex, race/ethnicity, etiology of ESRD, pretransplant diabetes, pretransplant time on dialysis, days on the waitlist, and percent panel-reactive antibody (PRA). Transplant procedure-associated covariates included transplant year, induction immunosuppression (lymphodepleting [antithymocyte globulin and alemtuzumab] vs nondepleting [basiliximab and daclizumab]) and maintenance immunosuppression regimen at discharge from the index hospitalization.

Propensity Score Matching

We used propensity score matching to balance important baseline characteristics between transplant recipients with ESRD due to amyloidosis and all other transplant recipients. We generated the propensity scores using logistic regression with several key covariates that were determined a priori (above and Tables S1-S3, SDC, We applied a nearest neighbor matching algorithm using a caliper of 0.01 with common support and no replacement to create 1:1 matches between the amyloidosis and “all other etiologies” exposure groups.25,26 In our matched analysis, we assessed for balance and bias using t testing for equality of the means in the 2 groups, standardized difference (ie, standardized percentage bias, reported as % bias) between the 2 groups, the variance ratio between the 2 groups (for continuous covariates), visual examination of histograms of propensity scores between the 2 exposure groups, as well as evaluation of Rubin’s B (reference range, <25%) and Rubin’s R (reference range, 0.5-2).27,28 After performing the propensity score matching, Cox proportional hazards regression was used to estimate hazard ratio (HRs) and 95% confidence intervals (CIs) for mortality and all-cause allograft failure. Robust sandwich estimation of the variance of the regression coefficient was used to account for clustering within the matched groups.29,30 The proportional hazards assumption was assessed via weighted versions of Kaplan-Meier curves using log-log plots as well as statistical testing and graphical displays based on the Schoenfeld and scaled Schoenfeld residuals.31

Secondary Analyses

Because of detection of effect modification by diabetes status and age, additional analyses were performed comparing patients with amyloidosis to those with ESRD due to diabetes and to those older than 65 years at the time of transplantation (elderly). For each of the propensity score-matched models, we also performed subgroup analysis in which we restricted the cohort to patients transplanted on or after 2001, often considered the “modern” era of transplantation medicine (with a shift towards tacrolimus-based maintenance immunosuppression) and AL amyloidosis therapy (with the routine use of autologous stem cell transplantation and new anti-plasma cell agents).

In secondary analyses, we also modeled the primary study outcomes, time to death, and time to all-cause allograft loss, using Cox proportional hazards regression.32 Any variable known to be a risk factor for patient or allograft loss as well as those associated with the exposure and outcome were included in the multivariable analysis. Confounders were retained in adjusted models if their inclusion changed the unadjusted HR of the outcome of interest in our exposure categories by more than 10% or were proposed a priori (recipient age, race, and sex).33 Model checking procedures, inclusive of examination of the proportional hazards assumption of the primary exposure and other covariates in multivariable adjusted models and visualization of log-log plots, were conducted for multivariable models generated for each outcome. If a variable was found to violate the proportional hazards assumption, that is, a P value less than 0.05 was found, but appeared to have a parallel appearance on visual inspection of the log-log plot, it was retained in the model without adjustment because this discrepancy was attributed to the large size of the data set.31

Statistical Analysis

Statistical analyses were performed using STATA version 14.0 (Statacorp LP, College Station, TX) with 2-sided hypothesis testing and a P value less than 0.05 as the criteria for statistical significance. Descriptive statistics (means, medians, and proportions) were used to describe baseline donor and recipient clinical and demographic characteristics comparing patients with and without ESRD attributed to amyloidosis. Continuous variables were compared using Student t test, or rank-sum test for abnormally distributed variables. Categorical variables were compared using χ2 test. Kaplan-Meier curves were generated and compared using log rank statistics.

Missing Data

Missing data were addressed with complete case analysis.34 Most covariates were missing in less than 5% of patients. Sensitivity analyses of the allograft failure outcome models were performed in variables that were greater than 5% missing (ie, PRA and calcineurin inhibitor immunosuppression), in which these variables were omitted to optimize patient sample size and to assess for any influence of missingness of these covariates in the overall outcomes. Additionally, recipient diabetes status was missing in 17% of patients with amyloidosis in the overall cohort; however, it was missing in less than 1% of patients transplanted on or after 2001; the 2001 subgroup analyses were used to corroborate the results considering the substantial reduction in missing data during that era.34,35 There was no information in the data set indicating whether the amyloidosis involved other organs in addition to the kidneys.


Patient Characteristics

We identified 310 629 adult, first kidney transplant recipients, of whom 576 were reported to have amyloidosis-associated ESRD. Demographic characteristics are presented in Table 1. Amyloidosis patients were older (median age, 57 vs 50 years; P < 0.001), were more likely to be white (81% vs 55.6%, P < 0.001), spent a shorter time on the waitlist (median, days 187 vs 393; P < 0.001), and were less likely to have diabetes (4% vs. 26%, P < 0.001). A smaller percentage of amyloidosis patients than nonamyloidosis patients received a deceased donor transplant (52% vs 64%, P < 0.001). Induction immunosuppression was more common in the amyloidosis cohort (67.5% vs 62.3%, P = 0.012), but maintenance immunosuppression was similar in both groups, with a majority of patients receiving a tacrolimus-based regimen at discharge from their index hospitalization. There was no statistically significant increase in rates of transplantation in patients with amyloidosis over the past 3 decades (0.17% of patients who underwent transplantation from 1987 to 1990 had ESRD due to amyloidosis, and 0.22% of patients who underwent transplantation from 2011 to 2015 had ESRD due to amyloidosis, P = 0.103, Figure 1). Clinical and demographic characteristics of patients transplanted in the modern era (2011-onward) are presented in Table S5 (SDC,

Clinical and demographic characteristics of the patient cohort, stratified by etiology of ESRD: amyloidosis versus nonamyloidosis
Proportion of patients with amyloidosis-associated ESRD among all kidney transplant recipients from 1987 to 2015.

Propensity Score Matching

We assembled a propensity score-matched cohort to compare amyloidosis patients to recipients with all other etiologies of ESRD, using 1:1 matching. There were no statistically significant differences noted between the 2 assembled cohorts in any of the matching variables (Table S1, SDC,

In our primary propensity score-matched cohort (481 amyloidosis to 481 nonamyloidosis), ESRD attributed to amyloidosis was associated with an increased risk of death compared to nonamyloidosis ESRD (HR, 1.58; 95% CI, 1.28-1.95, P < 0.001, Table 2A). The risk of allograft loss (395 amyloid matched 1:1 to 395 nonamyloid) was also increased in patients with ESRD due to amyloidosis (HR, 1.49; 95% CI, 1.19-1.87; P = 0.001) compared with their propensity score-matched counterparts, and this effect was independent of transplant era. In subgroup analyses restricting the cohort to patients transplanted on or after 2001, we found a similar pattern with increased risks of death (HR, 1.72; 95% CI, 1.31-2.26; P < 0.001) and graft loss (HR, 1.77; 95% CI, 1.35-2.33; P < 0.001, Table 2B) in the amyloidosis group compared with the nonamyloidosis group. Unadjusted Kaplan Meier survival curves are presented in Figures 2A and B.

Propensity score matched Cox regression models for patient mortality and allograft loss in the overall cohort
Propensity score-matched Cox regression models for patient mortality and allograft loss in patients transplanted on or after 2001
Patient and allograft survival in the amyloidosis versus nonamyloidosis cohorts.

We then performed analyses comparing recipients with ESRD due to amyloidosis to other high-risk groups, specifically patients with ESRD attributed to diabetes and elderly transplant recipients (Table 2A and 2B). In our propensity score-matched analysis comparing patients with ESRD due to amyloidosis versus diabetes, the risk of patient death was similar (HR, 0.99; 95% CI, 0.84-1.17; P = 0.902) as was the risk of allograft loss (HR, 1.00; 95% CI, 0.84-1.20; P = 0.975) (Table 2A). We found similar results when we compared amyloidosis patients with elderly recipients (mortality: HR, 0.99; 95% CI, 0.81-1.21; P = 0.910; graft loss: HR, 1.02; 95% CI, 0.82-1.26; P = 0.883). These results were unchanged in subgroup analyses of patients transplanted on or after 2001 (Table 2B). Unadjusted Kaplan Meier survival curves are presented in Figures 3 and 4. Additional sensitivity analyses in which we matched 1:5 instead of 1:1 yielded similar results to the overall cohort (Tables 2A and 2B).

Patient and allograft survival in the amyloidosis cohort compared to patients with ESRD attributed to DM.
Patient and allograft survival in the amyloidosis cohort compared to patients older than 65 years at the time of transplantation.

In secondary analyses using multivariable Cox regression rather than propensity matching, we again found an increased mortality risk associated with amyloidosis (HR, 1.85; 95% CI, 1.60-2.13; P < 0.001) in a model that included categorical age, sex, race, pretransplant dialysis exposure, recipient diabetes, donor type, KDPI and induction (Table 3; univariate analysis presented in Table S4, SDC, Similarly, amyloidosis was associated with an increased risk of allograft loss (HR, 1.62; 95% CI, 1.38-1.89; P < 0.001), in a model that incorporated categorical age, sex, race, pretransplant dialysis exposure, recipient diabetes, donor type, KDPI, cold ischemia time, induction, and maintenance immunosuppression (Table 3). Inclusion of transplant era did not significantly alter either model.

Multivariable Cox model for patient mortality and allograft loss

An etiology of death was only available for 62% of patients; the most commonly reported causes of death in both cohorts were cardiovascular disease (amyloidosis, 20.9% vs nonamyloidosis, 24.6%), infection (amyloidosis, 13.6% vs nonamyloid, 11.9%), and cancer (amyloidosis, 6.8% vs nonamyloidosis, 9.1%; P = 0.317). Of the 576 patients with amyloidosis-associated ESRD, 16 (2.7%) patients were reported to have died from amyloidosis. Similarly, information regarding the cause of allograft loss was only reported in 48% of patients. The most common cause of graft loss in both groups was rejection (amyloidosis, 37% and nonamyloidosis, 57.5%, P < 0.001). There were more graft losses due to “recurrent disease” in the amyloid cohort (15.7% vs 5.2%, P < 0.001); however, only 2 patients had recurrent amyloidosis specified as the etiology of graft failure.


In this study, we present the results of our retrospective, propensity score-matched cohort analysis assessing posttransplant outcomes for renal transplant recipients with ESRD due to amyloidosis. We demonstrated that although amyloidosis was associated with an increased risk of patient mortality and allograft loss, outcomes were like those observed in patients with diabetes and elderly patients, other high-risk patient subgroups that routinely undergo kidney transplant.

There are several small, mostly single center reports that have previously assessed outcomes for recipients with amyloidosis-associated ESRD.18-24 Most of these studies show inferior outcomes compared with the general population, with some suggesting that AA amyloidosis patients perform better than AL patients, and that regardless of the type of therapy used, AL patients who achieve partial or complete hematologic response to therapy pretransplant enjoy better outcomes than those who did not.18-24

In our cohort, the 1- and 5-year patient survival rates (91% and 70%, respectively) are like the results reported in the 1984 to 2009 UK study of AL amyloidosis patients (95% and 67%, respectively) and is better than previously reported US registry data (1-year survival of 87%).19,21 However, these outcomes are worse than the reported 5-year survival rate of 82% in AA amyloidosis patients from France.24 The median patient (5.8 years) and graft (4.8 years) survival rates in our cohort are like previously reported rates in AL amyloidosis patients but this median graft survival is lower than that reported from AA amyloid patients (10.3 years).22 Because the majority of kidney disease patients with amyloidosis in the United States have AL amyloidosis, we assume that this holds true of our overall UNOS cohort.5 If this assumption is correct, it would explain the similarities of our outcomes with previous reports of AL amyloidosis patients.

It is interesting that despite advances in solid organ transplantation, treatments for AL amyloidosis, and therapies for diseases underlying AA amyloidosis, outcomes were similar for patients transplanted before 2001 and those transplanted on or after 2001. Before the mid-late 1990s, the median survival after diagnosis for patients with AL amyloidosis was approximately 8 months. Therefore, we hypothesize that the proportion of AA/AL amyloidosis patients is different in the earlier era compared with the more recent era with more AA amyloidosis patients transplanted before 2001 and more AL amyloidosis patients on or after 2001. Because AL amyloidosis patients have worse outcomes than AA amyloidosis patients, this may have potentially masked the improved outcomes we expected to see with advances in medicine in the recent era.

Our study has important strengths. This analysis of kidney transplant recipients from 1987 to 2015 represents one of the largest published studies of postkidney transplant outcomes of patients with amyloidosis-associated ESRD in the United States. To our knowledge, comparisons of transplant outcomes between amyloidosis patients and other high risk recipient groups, such as diabetics and the elderly, have not been previously performed. The use of registry data minimizes the effects of center level and regional variations and allows for analysis of trends in patient and allograft survival that would otherwise be difficult for rare diseases. Our use of propensity score matching as our main analytical approach enables us to assess the effect of amyloidosis on posttransplant outcomes while accounting for selection bias and confounding by indication.

Although the use of registry data allowed us to report on a large cohort of patients with a rare disease, our study has several limitations. First, the type of amyloidosis is not collected by UNOS. Based on the epidemiology of the disease, the majority of US amyloidosis patients are likely to have AL disease, but because of the limited efficacy of antiplasma cell therapies before the mid-to-late 1990s, the relative proportions of AL and AA amyloidosis among the transplant recipients probably differed between the earlier and later periods of our study. This issue is important because Pinney et al22 found graft survival to be better in AA amyloidosis than AL amyloidosis. Second, data are not available regarding amyloidosis treatment before kidney transplant, response to treatment, or disease status at the time of transplant, each of which would be expected to affect posttransplant outcomes.21,23 Third, details regarding the extent of extrarenal amyloid, which can impact patient outcomes, are not available. Fourth, as with many studies using registry data, we are limited by the completeness and level of detail in the data set; for example, specific causes of death and graft loss were missing in nearly half of the patients. Additionally, we are unable to comment on recurrent amyloid after transplant or hematologic relapses after transplantation as these data are not collected in the UNOS data set. Also, there is a statistically shorter follow-up time for the amyloid cohort compared to the nonamyloid cohort. This may be a function of the decreased patient/allograft survival in the amyloid cohort, and thus, lower proportion of amyloid patients available for follow-up, but, nevertheless may have affected our interpretation of the results and should be considered another limitation. Finally, because transplant recipients usually represent a highly selected group of patients who are healthier than the general ESRD population, extrapolation of the findings to amyloidosis-associated ESRD patients without kidney transplants is not appropriate.

Kidney transplantation in patients with ESRD from amyloidosis has been controversial, and our retrospective propensity score matched cohort analysis demonstrates that although these recipients are at increased risk for death and graft loss compared with the kidney transplant population as a whole, their outcomes are comparable to other high-risk subgroups. Although these data overall are encouraging, these risks should be considered at the time of evaluation for renal transplantation and patients counseled about their risk appropriately. Furthermore, certain subgroups of amyloid patients, such as those older than 60 years, with diabetes or accepting high KDPI kidney may be at even greater risk for poor outcomes. For appropriately selected patients, amyloidosis should not preclude kidney transplantation. However, future study is needed to determine whether outcomes differ based on type of amyloidosis, and to further define the impact of extrarenal amyloid, contemporary amyloidosis therapies, posttransplant disease recurrence and hematologic relapse on posttransplant outcomes.


1. Sipe JD, Benson MD, Buxbaum JN, et al. Nomenclature 2014: amyloid fibril proteins and clinical classification of the amyloidosis. Amyloid. 2014;21:221–224.
2. Nienhuis HL, Bijzet J, Hazenberg BP. The prevalence and management of systemic amyloidosis in western countries. Kidney Dis (Basel). 2016;2:10–19.
3. Sanchorawala V. Light-chain (AL) amyloidosis: diagnosis and treatment. Clin J Am Soc Nephrol. 2006;1:1331–1341.
4. Westermark P, Westermark GT. Review. Reflections on amyloidosis in Papua New Guinea. Philos Trans R Soc Lond B Biol Sci. 2008;363:3701–3705.
5. Said SM, Valeri AM, Leung N, et al. Renal amyloidosis: origin and clinicopathologic correlations of 474 recent cases. Clin J Am Soc Nephrol. 2013;8:1515–1523.
6. Dember LM. Amyloidosis-associated kidney disease. J Am Soc Nephrol. 2006;17:3458–3471.
7. Gertz MA, Kyle RA, O’Fallon WM. Dialysis support of patients with primary systemic amyloidosis. A study of 211 patients. Arch Intern Med. 1992;152:2245–2250.
8. Kyle RA, Greipp PR, O'Fallon WM. Primary systemic amyloidosis: multivariate analysis for prognostic factors in 168 cases. Blood. 1986;68:220–224.
9. Palladini G, Hegenbart U, Milani P, et al. A staging system for renal outcome and early markers of renal response to chemotherapy in AL amyloidosis. Blood. 2014;124:2325–2332.
10. Nuvolone M, Merlini G. Systemic amyloidosis: novel therapies and role of biomarkers. Nephrol Dial Transplant. 2017;32:770–780.
11. Casserly LF, Fadia A, Sanchorawala V, et al. High dose intravenous melphalan with autologous stem cell transplantation in AL amyloidosis-associated end-stage renal disease. Kidney Int. 2003;63:1051–1057.
12. Muchtar E, Gertz MA, Kumar SK, et al. Improved outcomes for newly diagnosed AL amyloidosis between 2000 and 2014: cracking the glass ceiling of early death. Blood. 2017;129:2111–2119.
13. Lachmann HJ, Gallimore R, Gillmore JD, et al. Outcome in systemic AL amyloidosis in relation to changes in concentration of circulating free immunoglobulin light chains following chemotherapy. Br J Haematol. 2003;122:78–84.
14. Wechelaker AD, Gillmore JD, Hawkins PN. Systemic amyloidosis. Lancet. 2016;387:2641–2654.
15. D’Souza A, Dispenzieri A, Wirk B, et al. Improved outcomes after autologous hematopoietic cell transplantation for light chain amyloidosis: a center for international blood and marrow transplant research study. J CLin Oncol. 2015;33:3741–3749.
16. Lachmann HJ, Goodman HJ, Gilbertson JA, et al. Natural history and outcome in systemic AA amyloidosis. N Engl J Med. 2007;356:2361–2371.
17. Gillmore JD, Lovat LB, Persey MR, et al. Amyloid load and clinical outcome in AA amyloidosis in relation to circulating concentration of serum amyloid A protein. Lancet. 2001;358:24–29.
18. Pasternack A, Ahonen J, Kuhlbäck B. Renal transplantation in 45 patients with amyloidosis. Transplantation. 1986;42:598–601.
19. Bleyer AJ, Donaldson LA, McIntosh M, et al. Relationship between underlying renal disease and renal transplantation outcome. Am J Kidney Dis. 2001;37:1152–1161.
20. Sherif AM, Rafeie AF, Sobh MA, et al. Long-term outcome of live donor kidney transplantation for renal amyloidosis. Am J Kidney Dis. 2003;42:370–375.
21. Sattianayagam PT, Gibbs SD, Pinney JH, et al. Solid organ transplantation in AL amyloidosis. Am J Transplant. 2010;10:2124–2131.
22. Pinney JH, Lachmann HJ, Sattianayagam PT, et al. Renal transplantation in systemic amyloidosis—importance of amyloid fibril type and precursor protein abundance. Am J Transplant. 2013;13:433–441.
23. Herrmann SM, Gertz MA, Stegall MD, et al. Long-term outcomes of patients with light chain amyloidosis (AL) after renal transplantation with or without stem cell transplantation. Nephrol Dial Transplant. 2011;26:2032–2036.
24. Kofman T, Grimbert P, Canouï-Poitrine F, et al. Renal transplantation in patients with AA amyloidosis nephropathy: results from a French multicenter study. Am J Transplant. 2011;11:2423–2431.
25. Weitzen S, Lapane KL, Toledano AY, et al. Principles for modeling propensity scores in medical research: a systematic literature review. Pharmacoepidemiol Drug Saf. 2004;13:841–853.
26. Baser O. Too much ado about propensity score models? Comparing methods of propensity score matching. Value Health. 2006;9:377–385.
27. Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28:3083–3107.
28. Rubin DB. Using propensity scores to help design observational studies: application to the tobacco litigation. Health Serv Outcomes Res Methodol. 2001;2:169–188.
29. Austin PC. The performance of different propensity score methods for estimating marginal hazard ratios. Stat Med. 2013;32:2837–2849.
30. Yin DY, Wei LJ. The robust inference for the Cox proportional hazards model. J Am Stat Assoc. 1989;84:1074–1078.
31. Therneau TM, Grambsch PM. Modeling survival data: extending the cox model. New York, NY: Springer; 2001.
32. Cox D. Regression models and life tables. J R Stat Soc Ser. 1972;34:187–220.
33. Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia, USA: Lippincott Williams and Wilkins; 1998.
34. White IR, Carline JB. Bias and efficiency of multiple imputation compared with complete-case analysis for missing covariate values. Stat Med. 2010;29:2920–2931.
35. Bartlett JW, Carpenter JR, Tilling K, et al. Improving upon the efficiency of complete case analysis when covariates are MNAR. Biostatistics. 2014;15:719–730.
36. Rao PS, Schaubel DE, Guidinger MK, et al. A comprehensive risk quantification score for deceased donor kidneys: the kidney donor risk index. Transplantation. 2009;88:231–236.
37. Cohen JB, Bloom RD, Reese PP, et al. National outcomes of kidney transplantation from deceased diabetic donors. Kidney Int. 2016;89:636–647.
38. Faravardeh A, Eickhoff M, Jackson S, et al. Predictors of graft failure and death in elderly kidney transplant recipients. Transplantation. 2013;96:1089–1096.
39. Narayanan M, Pankewycz O, Shihab F, et al. Long-term outcomes in African American kidney transplant recipients under contemporary immunosuppression: a four-yr analysis of the Mycophenolic acid Observational REnal transplant (MORE) study. Clin Transplant. 2014;28:184–191.

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.