“Nature versus Nurture” Study of Deceased-Donor Pairs in Kidney Transplantation

Abstract

Donor characteristics such as age and cause of death influence the incidence of delayed graft function (DGF) and graft survival; however, the relative influence of donor characteristics (“nature”) versus transplant center characteristics (“nurture”) on deceased-donor kidney transplant outcomes is unknown. We examined the risks for DGF and allograft failure within 19,461 recipient pairs of the same donor's kidneys using data from the US Renal Data System. For the 11,894 common-donor pairs transplanted at different centers, a recipient was twice as likely to develop DGF when the recipient of the contralateral kidney developed DGF (odds ratio [OR] 2.05; 95% confidence interval [CI] 1.82 to 2.30). Similarly, for 7567 common-donor pairs transplanted at the same center, the OR for DGF was 3.02 (95% CI 2.62 to 3.48). For pairs transplanted at the same center, there was an additional 42% risk for DGF compared with pairs transplanted at different centers. After adjustment for DGF, the within-pair ORs for allograft failure by 1 yr were 1.92 (95% CI 1.33 to 2.77) and 1.77 (95% CI 1.25 to 2.52) for recipients who underwent transplantation at the same center and different centers, respectively. These data suggest that both unmeasured donor characteristics and transplant center characteristics contribute to the risk for DGF and that the former also contribute significantly to allograft failure.

Delayed graft function (DGF) is an important predictor of graft failure after kidney transplantation.^1–3 The incidence of DGF after deceased-donor kidney transplants ranges between 23 and 50%.^4–6 Although some studies have been mixed, several large studies have shown that DGF influences graft failure both through its association with and independent of acute rejection.^5,7–10 DGF also adversely affects cost, length of hospitalization, and patient rehabilitation.^11–13 Allograft failure results in half of the deceased-donor kidneys being lost at 11 yr after transplantation.¹⁴

There are many known determinants of DGF and allograft failure. Studies have implicated a number of immunologic and nonimmunologic characteristics, including donor factors, recipient factors, and the transplant procedure.^4,6,15–21 A limited effort has been made to evaluate the relative contribution of these risk factors by exploiting that there is variation in the response of recipients of kidneys from the same donor.^18,22–24 This approach is similar to studies of monozygotic twins reared apart, which seek to quantify the relative importance of environmental and genetic factors on the basis of variability within twin pairs and among twin pairs.^22,25 Analyses that examine outcomes in two recipients of kidneys from the same deceased donor can be used to determine the donor's relative contribution to the recipients’ outcomes.

We retrospectively evaluated a national cohort of deceased-donor transplant recipients to understand better the complex relationship between donor (“nature”) and transplant center effects (“nurture”) associated with DGF and kidney allograft failure. We examined the within-pair correlation of these outcomes among recipients of kidneys from the same deceased donor and adjusted for transplant center effect by estimating separate odds ratios (ORs) for recipient pairs who underwent transplantation at the same transplant center and at different transplant centers. The transplant center effect was detected by determining the difference in outcomes for the paired kidneys from the same deceased donor transplanted at the same versus different centers.

RESULTS

Delayed Graft Function

DGF occurred in 9430 (24%) kidney transplant recipients (Tables 1 and 2). A total of 1846 (10%) common-donor recipient pairs were concordant with both pairs’ developing DGF, and 11,877 (61%) pairs were concordant with both pairs’ not developing DGF. In contrast, only 5738 (29%) pairs were discordant for DGF. Within pairs of recipients with a common donor, when one recipient experienced DGF, the unadjusted OR for DGF in the recipient of the contralateral kidney was 2.66 (95% confidence interval [CI] 2.48 to 2.86; P < 0.001). In the adjusted model (adjusted for variables listed in Table 3), when one recipient experienced DGF, the recipient of the contralateral kidney was at increased risk for DGF (both recipients in the same center: OR 3.02 [95% CI 2.62 to 3.48; P < 0.001]; different center: OR 2.05 [95% CI 1.82 to 2.30; P < 0.001]). Expanded-criteria donor (ECD) was a potent risk factor for DGF (OR 1.51; 95% CI 1.39 to 1.63; P < 0.001; Table 3). In the model limited to recipients with Centers for Medicare and Medicaid Services (CMS) claims for dialysis after transplantation, when one recipient experienced DGF, the recipient of the contralateral kidney was still at increased risk for DGF (both recipients at the same center: adjusted OR 2.76 [95% CI 2.23 to 3.42; P < 0.001]; different centers: OR 2.26 [95% CI 1.89 to 2.71; P < 0.001]). Including only recipients with United Network for Organ Sharing (UNOS) form data yielded a similar OR for DGF (both recipients in the same center: OR 3.73 [95% CI 3.22 to 4.32; P < 0.001]; different center: OR 1.97 [95% CI 1.74 to 2.23; P < 0.001]). We conducted a sensitivity analysis by including the 4075 recipients who had allograft failure within 3 mo of transplantation to determine whether primary nonfunction affected our findings, and the results were not statistically different. DGF was independently associated with kidney allograft failure (hazard ratio 1.30; 95% CI 1.22 to 1.38; P < 0.001). Table 4 lists an adjusted estimation of the proportion of recipients’ DGF risk attributable to DGF in the recipient of the contralateral kidney or to an ECD, a potent risk factor for DGF. Considering DGF in the recipient of the contralateral kidney as a surrogate for exposure to unmeasured donor factors, these unmeasured factors contribute more to a recipient's risk for DGF than does exposure to an ECD alone (17 versus 5%), even after accounting for the center effect.

Table 1.

Characteristics of all donors, donors whose recipients underwent transplantation at different centers, and donors whose recipients underwent transplantation at the same centers^a

Donor Characteristic	All Donors (%; n = 19,461)	Transplantation at Different Centers (%; n = 11,894)	Transplantation at the Same Center (%; n = 7567)
Age (yr)
<18	18	19	17
18 to 34	27	28	26
35 to 49	30	30	30
50 to 64	22	20	24
≥65	3	2	3
Male	59	60	58
Race
white	83	84	82
black	11	10	12
Native American	<1	<1	<1
Asian	2	1	2
other	4	4	4
ECD kidney	16	14	18
Terminal creatinine >1.5 mg/dl	12	11	14
Death from CVA	55	54	56
Death from trauma	52	53	50
Donor hypertension	19	18	21

^aCVA, cerebrovascular accident.

Table 2.

Recipient and transplantation characteristics

Characteristic	Value(n = 38,922)
Age (yr; %)
<18	16
18 to 34	34
35 to 49	37
50 to 64	9
Male (%)	61
Race/ethnicity (%)
white	63
black	30
Native American	1
Asian	5
other	<1
Primary native kidney disease (%)
diabetes	24
hypertension	22
glomerulonephritis	25
other	29
Hepatitis C seropositive (%)	6
Peak PRA ≥10 (%)	22
Zero HLA mismatches (%)	12
DGF (%)	24
Medicare as primary payor (%)	57
Dialysis time (yr; %)
0 (preemptive)	6
<1.0	14
1.0 to 2.9	41
≥3.0	40
No. of HLA mismatches (%)
0	12
1 to 3	33
4 to 6	55
Cold ischemia time (h; %)
<12	20
12 to 24	56
>24	24
Mechanical perfusion (%)	12
Preservation solution (%)
University of Wisconsin solution	72
other	2
unknown	25
Induction therapy (%)
IL-2 antibody	25
none	49
other	26
Mean no. of transplants at center per year, 1995 to 2003 (%)
≤20	7
20 to 60	32
>60	61
Median distance organ traveled from procurement hospital to transplant center (mi; first to third quartiles)	68 (6 to 272)

^aPRA, panel reactive antigen.

Table 3.

Odds of DGF given that the recipient of the contralateral donor kidney had DGF^a

Variable (Reference Group)	OR	95% CI	P
Recipient age (versus <18 yr)
18 to 34	1.13	0.93 to 1.39	0.2200
35 to 49	1.18	0.97 to 1.43	0.1000
50 to 64	1.19	0.98 to 1.45	0.0700
≥65	1.20	0.98 to 1.48	0.0800
Recipient race (versus white)
Asian	0.92	0.82 to 1.05	0.2100
black	1.32	1.24 to 1.40	<0.0001
Native American	1.08	0.86 to 1.35	0.5200
other	1.25	0.80 to 1.96	0.3200
Recipient gender (male versus female)	1.30	1.23 to 1.38	<0.0010
Primary native kidney disease (versus diabetes)
other	0.68	0.63 to 0.74	<0.0010
hypertension	0.76	0.70 to 0.82	<0.0010
glomerulonephritis	0.73	0.68 to 0.79	<0.0010
Hepatitis C seropositive recipient (yes versus no)	1.05	0.94 to 1.18	0.3900
Medicare as primary payor (yes versus no)	1.29	1.22 to 1.37	<0.0010
Peak PRA ≥10 (versus <10%)	1.19	1.12 to 1.27	<0.0010
HLA mismatches (versus 0)
1 to 3	1.14	1.04 to 1.26	0.0070
4 to 6	1.28	1.17 to 1.41	<0.0001
Time on dialysis before transplantation (versus none)
0.0 to 0.9 yr	2.56	2.05 to 3.19	<0.0001
1.0 to 2.9 yr	3.76	3.04 to 4.64	<0.0001
≥3.0 yr	5.23	4.23 to 6.48	<0.0001
Donor age (versus <18 yr)
18 to 34 yr	1.24	1.13 to 1.36	<0.0001
35 to 49 yr	1.66	1.51 to 1.83	<0.0001
50 to 64 yr	1.95	1.76 to 2.17	<0.0001
≥65 yr	2.16	1.81 to 2.59	<0.0001
Donor race (versus white)
Asian	0.92	0.74 to 1.14	0.4400
black	0.92	0.83 to 1.01	0.0800
Native American	1.13	0.69 to 1.88	0.6200
other	1.07	0.92 to 1.25	0.3500
Donor male (versus female)	1.04	0.98 to 1.11	0.2100
Terminal creatinine >1.5 mg/dl (versus <1.5 mg/dl)	1.71	1.57 to 1.86	<0.0001
Donor hypertension (yes versus no)	1.32	1.23 to 1.43	<0.0010
Donor cause of death CVA (yes versus no)	1.20	1.08 to 1.33	0.0005
Donor cause of death trauma (yes versus no)	0.98	0.89 to 1.09	0.7400
Preservation solution (versus University of Wisconsin solution)
other	1.07	0.88 to 1.31	0.4900
unknown	0.89	0.82 to 0.95	0.0011
Mechanical perfusion (versus none)	0.56	0.51 to 0.62	<0.0010
Cold ischemia time (versus <12 h)b
12 to 24 h	1.59	1.47 to 1.71	<0.0001
>24 h	2.34	2.14 to 2.55	<0.0001
>36 h	4.03	3.45 to 4.70	<0.0001
Mean no. of transplants at center per year, 1995 to 2003 (versus ≤20 per year)
20 to 60	1.03	0.92 to 1.16	0.6100
>60	0.92	0.82 to 1.03	0.1700

^aResults of logistic regression with ORs for DGF in recipients of a donor pair, given that the outcome occurred in the recipient of the contralateral kidney, with both recipients undergoing transplantation at the same center or at different centers. This model was also adjusted for initial immunosuppression medication and induction therapy (data not shown). ECD defined as donor age >60 yr or as donor age >50 yr with any two of the following donor criteria: (1) Terminal serum creatinine >1.5 mg/dl, (2) hypertension, or (3) death from CVA. The model in this table, with ECD instead of the component variables, showed a strong association between ECD status and DGF (OR 1.51; 95% CI 1.39 to 1.63; P < 0.001).

^bCold ischemia time was highly correlated with distance organ traveled from procurement hospital to transplant center; therefore, only cold ischemia time was included in this model.

Table 4.

Risk for DGF attributable to unmeasured donor factors in the common-donor recipient pair and to ECD status^a

Risk Factor	Transplant Location	Attributable Risk (95% CI)
DGF in the recipient of the contralateral kidney	All centers	0.17 (0.16 to 0.20)
	Different center	0.15 (0.13 to 0.18)
	Same centers	0.26 (0.22 to 0.29)
ECD status of the transplanted kidney	All centers	0.05 (0.04 to 0.07)
	Different center	0.04 (0.03 to 0.07)
	Same centers	0.06 (0.04 to 0.09)

^aAttributable risk estimates adjusted for variables associated with DGF (as shown in Table 3), such as Medicare primary payor status (whether the DGF information came from Medicare claims or Organ Procurement and Transplantation Network forms), recipient race, recipient gender, primary native kidney disease, PRA, time on dialysis before transplantation, HLA mismatches, cold ischemia time and mechanical perfusion.

Serum Creatinine at Hospital Discharge

Given the correlation of DGF within common-donor pairs, we also explored a correlation of early kidney function. Using the hospital discharge serum creatinine as a measure of early kidney allograft function, we again observed significant correlation within recipients with a common donor. In an adjusted model (adjusted for variables listed in Table 3), there was an adjusted correlation of 0.38 and 0.33 in discharge creatinine within recipient pairs at the same and different centers, respectively (linear mixed effects model, P < 0.001). This correlation is higher than the correlation of 0.29 reported for serum creatinine between monozygotic twins in the literature.²⁶

Kidney Allograft Failure

Using 1-, 2-, and 3-yr death-censored allograft failure and allograft failure as binary outcomes, there existed in the adjusted model a correlation within recipients of kidneys from the same deceased donor (Table 5). The ORs for allograft failure within pairs of recipients with the same deceased donor were similar in magnitude when the recipients underwent transplantation at the same center versus different centers; therefore, no significant transplant center effect was detected after adjustment for DGF. At year 1, the increase in risk for allograft failure for patients who underwent transplantation at the same center was not statistically significant (OR 1.57; 95% CI 0.65 to 3.78) compared with patients who underwent transplantation at different centers. The correlation within recipients of kidneys from the same donor attenuated with time, suggesting that unmeasured recipient comorbidities may play an increasing role in later outcomes. In particular, the within-pair correlation for allograft failure remained highly significant even at 3 yr after transplantation. We noted no era effect in the outcomes at same or different centers (data not shown). We conducted a sensitivity analysis by including the 4075 recipients who had allograft failure within 3 mo of transplantation to determine whether primary nonfunction affected our findings, and the results were not statistically different (data not shown).

Table 5.

Outcomes of common-donor recipient pairs who underwent transplantation at the same or at different transplant centers^a

Posttransplantation Year	Death-Censored Allograft Failure (OR [95% CI])	Allograft Failure (OR [95% CI])
Year 1
same center	2.76 (1.57 to 4.87)	1.92 (1.33 to 2.77)
different center	1.69 (0.87 to 3.31)	1.77 (1.25 to 2.52)
Year 2
same center	1.90 (1.35 to 2.69)	1.53 (1.23 to 1.91)
different center	1.60 (1.15 to 2.24)	1.44 (1.18 to 1.76)
Year 3
same center	1.81 (1.38 to 2.39)	1.43 (1.18 to 1.73)
different center	1.83 (1.43 to 2.35)	1.50 (1.28 to 1.76)

^aResults of logistic regression with ORs for death-censored allograft failure and allograft failure in recipients of a donor pair, given that the outcome occurred in the recipient of the contralateral kidney, with both recipients undergoing transplantation at the same center or at different centers. All ORs were adjusted for variables listed in Table 3, including DGF, and were significant at P < 0.001.

DISCUSSION

This study has two main findings. First, there was a significant correlation within pairs of kidney transplants from the same donor for the occurrence of DGF and allograft failure, suggesting that unmeasured donor factors (“nature”) contribute to transplant outcomes. These unmeasured donor factors had a higher attributable risk for DGF than an ECD, a potent risk factor for DGF. Second, there was a transplant center effect (“nurture”) on DGF but not on allograft failure within the first 3 yr after transplantation. We observed that within pairs of recipients with a common donor, when DGF occurred in one recipient, the adjusted odds for DGF in the recipient of the contralateral kidney was >200% higher when recipients underwent transplantation at the same center and >100% higher when recipients underwent transplantation at different centers. Considering DGF in the recipient of the contralateral kidney as a surrogate for exposure to unmeasured donor factors, these factors together contribute more to a recipient's risk for DGF than does exposure to an ECD (17 versus 5%), even after accounting for the center effect (Table 4). Evidence that unmeasured donor factors contribute to early allograft dysfunction is also shown by a correlation within pairs for serum creatinine at hospital discharge. After adjustment, 38 and 33% of the correlation in discharge creatinine was evident within recipient pairs who underwent transplantation at the same and different centers, respectively. Similarly, the risk for allograft failure for a recipient when the recipient of the contralateral kidney also experienced allograft failure was >50% higher, even at 3 yr after transplantation.

Similarities in the outcomes of DGF and allograft failure within common-donor recipient pairs are due, in part, to similarities in the posttransplantation treatment of patients at transplant centers. This is reflected by the higher ORs when both recipients underwent transplantation at the same center versus different centers (Table 5). Interestingly, we could not detect a transplant center effect on the correlation for kidney allograft failure in the first 3 yr after transplantation. This reflects more variation in early posttransplantation treatment than later treatment by the transplant center.

The correlations for DGF and allograft failure within common-donor recipient pairs indicate that donor variables play an important role in kidney transplant function and allograft failure. Our results are consistent with previous, smaller studies.^22–24 In those previous studies, no discernible effect of donor factors on allograft function was present at 6 mo after transplantation. In contrast, our study suggests that donor factors remain important contributors to allograft survival up to 3 yr after transplantation. Two studies of common-donor recipient pairs detected similar rates of DGF within each of the common-donor recipient pairs, similar to our study.^23,24

Our study suggests that unmeasured donor factors such as donor genetic variants may play a significant role in transplant outcomes. There is an emerging interest in the possibility that genetic or biochemical variation among donors plays a role in kidney transplant outcomes.^27,28 It was recently suggested that postreperfusion kidney biopsy transcriptomes or donor genotype can be used to identify donor kidneys at increased risk for DGF.^28,29

In a recent report, Knowing What Works in Healthcare, the Institute of Medicine highlighted that “variations in how health care providers treat specific conditions reflect uncertainty and disagreement about what the clinical standard should be” and that “spending on ineffective care … contributes to soaring health costs.”³⁰ The effects of transplant center on early and late allograft outcomes have been documented in the United States, Canada, England, and Ireland,^31–35 and adherence with immunosuppressants is associated with the transplant center.³⁶ Our study is the first to study the transplant center effect using common-donor recipients who underwent transplantation at the same and at different centers. Here, we showed that the within-pair odds for DGF were influenced by the transplant center. Because our analysis was of recipients of kidneys from the same donor, we were able to control for the influence of the donor on transplant outcomes as we determined the center effect. In the past, it was suggested that half of the center effect was associated with events that occurred during the transplant hospitalization.³⁷ Consistent with this, our study shows that the center effect on DGF is strong.

Our study has some notable limitations. First, there is the possibility for residual confounding as a result of clinical factors not included in the US Renal Data System (USRDS). Our analyses included many but not all of the factors used to make clinical decisions to confer risk in allograft selection and to guide posttransplantation treatment. Such factors not in the USRDS include donor histology at time of preimplantation and donor management factors such as donor hemodynamics before procurement. Second, DGF was defined as the need for dialysis in the first week after transplantation. DGF was not determined by using creatinine clearance to measure kidney function, because such detailed creatinine levels and exact date of creatinine test are not available in the USRDS. It is possible that some DGF center effects may be due to differences in threshold for using dialysis after transplantation. A subset analysis limited to patients with more than one dialysis treatment in the first week after transplantation, before hospital discharge, showed similar within-pair correlation of DGF (data not shown). Using creatinine clearance is also problematic in patients who undergo dialysis. Despite the potentially large measurement error in serum creatinine, we were still able to show a strong association between discharge creatinine within pairs of recipients with a common donor.

In conclusion, our study showed that early allograft dysfunction is strongly influenced by both donor and transplant center effects, and allograft failure is strongly influenced by donor effects. To our knowledge, this represents the largest study to date examining the correlation of outcomes in recipients of a kidney from a common donor and examining the transplant center effect. We are further investigating how donor genetic factors and how the variation in transplant center care (e.g., frequency of visits, type of provider seen) affect transplant outcomes. Because DGF and allograft failure have significant health and economic costs, donor factors represent important opportunities to improve renal transplant outcomes and provide the most effective care.

CONCISE METHODS

Study Population

There were 110,830 first-time kidney transplant recipients between December 31, 1994, and December 31, 2003, in the United States. To obtain a cohort of kidney transplant recipient pairs with a common deceased donor from the USRDS registry, we excluded living-donor recipients (40,193), recipients of en block or double kidney transplants (1985), simultaneous pancreas transplant recipients (7929), and recipients of kidneys from deceased donors whose contralateral kidneys were not transplanted in first-time kidney recipients (17,726). To remove patients with primary nonfunction of transplanted kidneys, we excluded an additional 4075 recipients who had allograft failure within 3 mo of transplantation. This left a final study cohort of 38,922 recipients, or 19,461 common-donor recipient pairs. We conducted a sensitivity analysis by including the 4075 recipients who had allograft failure within 3 mo of transplantation to determine whether primary nonfunction affected our findings.

Data Sources and Definitions

Data were from CMS and UNOS recorded in the USRDS registry. Patients with DGF were defined as needing dialysis treatment within the first 7 d after transplantation. Dialysis treatments were identified using CMS claims (available for Medicare primary payer patients). When CMS claims were not available, Organ Procurement and Transplantation Network data from UNOS were used. This strategy was validated in a previous study.³⁸ Serum creatinine was measured at hospital discharge for each recipient and used as an early index of kidney function. Death-censored allograft failure was defined as return to dialysis or retransplantation. Allograft failure was defined by return to dialysis, retransplantation, or death with a functioning kidney. Individuals not experiencing graft failure were censored at December 31, 2004, when the last follow-up date was available. Distance that an organ traveled from procurement hospital to transplant center was calculated using zip codes as provided by UNOS. All zip codes were matched to the ZIPCODE data file provided by the SAS Institute (Cary, NC) with SAS 9.1.3. This data file contains the latitude and longitude of the centroid of every US zip code. The distance between each organ's procurement hospital and transplant center was then estimated as the distance between the centroids of the zip codes using the great circle formula.³⁹

Statistical Analysis

Logistic regression was used to estimate an unadjusted OR to assess the crude correlation of DGF within recipient pairs. Generalized estimating equation estimation, which accounts for the correlation of DGF occurrence within recipient pairs using the alternating logistic regression (ALR) approach, was used to fit adjusted models with the measured donor, recipient, and procedural factors used as covariates.⁴⁰ ALR was used instead of a time-to-event survival analysis such as the Cox survival analysis because our primary interest was the within-pair risks for graft failure for common-donor recipients, rather than a comparison of average cumulative survival for the recipients of the right or left kidney. ALR allows for the estimation of pair-wise ORs of outcomes within a cluster (i.e., pairs of recipients with the same donor) and estimation of same and different transplant center effects while accounting for the dependence of the outcome on covariates.⁴¹ Given the significant correlation between cold ischemia time and distance the kidney traveled from the procurement to the transplant hospital, only cold ischemia time was included in the model. We evaluated the effect on within-pair risk for DGF when both common-donor recipients underwent transplantation at the same transplant center (7567 pairs) and when both recipients underwent transplantation at different centers (11,894 pairs). In a subset analysis, separate models using DGF defined using CMS claims only and Organ Procurement and Transplantation Network form data only were fit. The additional risk of transplants at the same center versus different centers was assessed by calculating the difference of the log ORs of the same center pairs and the different center pairs. The variance and thus the CIs were calculated on the covariance matrix of the two estimated effects.

We estimated attributable risks to compare the relative proportion of the risk for DGF experienced by recipients with (1) DGF in the recipient of the contralateral kidney or (2) an ECD. ECD was defined as donor age >60 or as donor age >50 with any two of the following donor criteria: (1) Donor serum creatinine at the time of organ procurement (terminal creatinine) >1.5 mg/dl, (2) hypertension, or (3) death from cerebrovascular accident. To calculate attributable risk, the DGF status of one of the pairs of recipients with a common donor was randomly selected as the outcome, and the DGF status of the recipient of the contralateral kidney was the exposure. The attributable risk estimates were adjusted for all factors that were significantly associated with DGF and either ECD status (for estimating risk attributable to DGF in the recipient of the contralateral kidney) or DGF in the recipient of the contralateral kidney (for estimating risk attributable to an ECD). The estimates were calculated for common-donor recipients who underwent transplantation at the same versus different centers. The adjusted attributable risks were calculated using R package pARtial (http://cran.r-project.org/web/packages/pARtial/index.html) using the model-free approach.⁴²

The relationship within pairs for serum creatinine (log transformed) at the time of hospital discharge was evaluated using a linear mixed model. The outcomes of death-censored allograft failure and allograft failure were analyzed with ALR models at 1, 2, and 3 yr after transplantation. The estimates were calculated for common-donor recipients who underwent transplantation at the same versus different centers. The year of transplantation was included in the ALR models to test for an era effect. A Cox proportional hazards model was used to determine whether DGF was associated with kidney allograft failure.

Statistical analyses were performed using SAS 9.1 (SAS Institute) and R 2.6.⁴³ Attributable risks in the analysis of DGF were estimated using R, package pARtial. P values were two-sided, and P < 0.05 was considered significant.

DISCLOSURES

None.