Expanded Criteria Donor Kidney Allocation: Marked Decrease in Cold Ischemia and Delayed Graft Function at a Single Center



Expanded criteria donor (ECD) kidney allocation aims to increase utilization and facilitate placement. We implemented an ECD program for pre-consented candidates and studied whether ECD allocation decreased cold ischemia time and delayed graft function (DGF). We compared donor, recipient and transplant data for ECD transplants performed during the first year of our program to those performed in the preceding 5½ years. Logistic regression identified risk factors for DGF. Of 356 candidates, 107 (30%) consented, 32 (9%) completed evaluation and 20 (6%) underwent ECD transplantation during the program's first year. The recent and historical ECD cohorts had similar donor and recipient characteristics, except that recent ECD recipients were older. The rate of donor kidney biopsy dropped from 85% to 24% (p < 0.001). Cold ischemia time decreased from 16.4 to 7.4 h (p < 0.001), as did the incidence of DGF from 43% to 15% (p = 0.031). Three independent risk factors for DGF emerged: recipient height (OR 1.21/10 cm; p = 0.008), >4 HLA mismatches (OR 20.46; p = 0.0033) and cold ischemia time (OR 1.24/h; p = 0.0036). We conclude the ECD designation provides a description of kidney quality that may obviate biopsy. ECD allocation decreased cold ischemia time and DGF, which may improve graft survival.


The major limitation facing kidney transplantation is a critical shortage of donor organs. Advances in surgical technique, peri-operative care, and immunosuppression have resulted in excellent transplant outcomes with minimal recipient morbidity and mortality. As a result, transplantation is now recognized as the best therapy for end-stage renal disease, providing survival benefit to essentially all wait-listed dialysis patients (1).

As demand for transplantation has grown, so has time spent on dialysis awaiting transplant. Wait time exerts a negative effect on transplant outcomes. Meier-Kriesche et al. showed that the duration of dialysis before transplant was the most important modifiable risk factor for graft and patient survival (2,3). Transplantation that pre-empted dialysis improved outcomes, whereas longer time on dialysis before transplant compromised outcomes. To offer transplantation to as many candidates as possible, as quickly as possible, the transplant community has aggressively pursued strategies to expand the kidney supply.

One approach has been to broaden the criteria for kidneys considered acceptable for transplantation. A young healthy person who suffered brain death from trauma has long been considered the ideal donor. In contrast, the term ‘non-ideal’ has connoted older donors, those with chronic conditions such as hypertension or diabetes, and/or those who donate after cardiac death. Although transplant outcomes of non-ideal kidneys are inferior to those of ideal kidneys, Ojo et al. showed that they offer significant survival benefit over continued dialysis (4).

Despite their survival benefit, analysis of national procurement and utilization patterns revealed that non-ideal kidneys have a high discard rate. For example, between 1996 and 2001, the overall kidney discard rate increased from 12% to 15%, a trend driven primarily by the nearly 50% discard rate of kidneys from donors older than 60 years (5–7). Furthermore, since non-ideal kidneys were often declined by one or more centers before eventual placement, they were frequently associated with prolonged cold ischemia time. Cold ischemia time exerts a disproportionate effect on kidneys from older donors: at a given cold ischemia time, older kidneys have a higher incidence of DGF than younger kidneys (5,8,9). These findings motivated the transplant community to develop a new allocation policy for non-ideal kidneys designed to enhance utilization and decrease cold ischemia time.

On October 31, 2002, the United Network for Organ Sharing (UNOS) launched a new policy specifically allocating kidneys from expanded criteria donors (ECDs). ECD kidneys were defined as kidneys with a relative risk of graft loss >1.7 compared to ideal kidneys, based upon four simple donor characteristics available at the time of procurement (5,6). The ideal kidney is one from a donor 18–40 years old with no medical co-morbidities. ECD kidneys include any donor over the age of 60 and any donor 50–59 years of age with at least two of three medical conditions: brain death from cerebrovascular accident, history of hypertension, or creatinine at procurement >1.5 mg/dL. The policy stipulated that ECD kidneys would be offered to candidates on a separate waiting list, aiming to minimize the number of offers required to place the kidney. To increase the predictability of ECD kidney offers, allocation was based on wait time only, except for zero HLA mismatch situations.

Since the ECD allocation policy was introduced, few reports have described whether the goals of increasing utilization and decreasing cold ischemia time have been achieved. Very recently, an analysis of national data showed that both ECD kidney recovery and transplantation have increased significantly since implementation of the ECD policy (10). More ECD kidney transplants had cold ischemia times <12 h while fewer had cold ischemia times >24 h (10). We aimed to explore the effect of ECD allocation at the single-center level. In this report, we describe the design and implementation of our ECD program. We compared prospectively collected data on ECD transplants performed over 1 year with data from a historical cohort of ECD transplants to determine the impact of the new ECD allocation policy on outcomes, with focus on cold ischemia time and delayed graft function.

Materials and Methods

Patient selection and evaluation

We used recommendations generated by a consensus conference to develop inclusion and exclusion criteria for our ECD program (7). We chose to offer ECD kidneys to recently listed candidates at high risk for mortality on dialysis who also had a lower requirement for nephron mass. Specifically, all wait-listed candidates older than 60 years and all diabetic candidates older than 40 years were identified as potential ECD candidates, excluding those who weighed >80 kg for men and >70 kg for women, those who were sensitized (prior transplant or PRA >10%), and those with additional wait time less than 2 years (Table 1). Since our organ procurement organization has a UNOS variance to allocate kidneys strictly based upon waiting time, without weight given to donor-recipient immunological matching (except for 0 mismatch sharing), timing of non-ECD kidney transplantation was highly predictable. As such, candidates with >2 years additional wait time were readily identifiable. Potential candidates and their referring nephrologists were sent a letter explaining the new ECD allocation policy and our new program. Toward the end of the study period, qualified patients were given information directly during their initial visit. Those willing to accept an ECD kidney were asked to return a signed and witnessed consent form. One transplant coordinator was designated to administer the program.

Table 1. Inclusion and exclusion criteria for the UCSF Expanded Criteria Donor program
Inclusion criteria:
 1. Any candidate older than 60 years
 2. Any diabetic candidate older than 40 years
Exclusion criteria:
 1. History of prior transplant (kidney or other organ)
 2. Panel reactive antibody >10%
 3. Large body mass (>80 kg for men or >70 kg for women)
 4. Anticipated to be within 2 years of transplantation
  with a non-ECD kidney

After receiving signed consent, one transplant surgeon reviewed each candidate's medical record to expedite full evaluation according to our center's protocols. Although all of these candidates had been accepted and wait-listed for transplantation, they were typically far from having undergone a complete transplant evaluation. At our center, referred candidates were accepted and wait-listed for transplantation based upon available medical records and information gathered from interviews by the transplant physician or surgeon, the nurse coordinator and the social worker during their initial visit to our center. Additional testing to complete the evaluation was not performed until 1 year before transplantation, a predictable time point because all kidneys were allocated based upon wait time alone in our donor service area. The return of a signed ECD consent form initiated testing required to complete transplant evaluation. Results were then presented to the multidisciplinary Kidney Transplant Selection Committee for approval. Once approved, the candidate's serum was placed on trays to enable crossmatching against future donors. Candidates were then transplanted when ECD kidneys became available.

Study design and data collection

We prospectively collected donor, recipient, and transplant data for 20 consecutive ECD transplants performed during the first year of our program (July 28, 2003–July 27, 2004). The same information was retrospectively collected for a historical cohort of 54 transplants performed in the preceding 5½ years (January 1, 1998–July 27, 2003) using kidneys from donors who would have met ECD criteria. This historical cohort served as controls to determine the impact of ECD allocation. While current ECD transplants had to meet inclusion/exclusion criteria, all wait-listed candidates were eligible for historical ECD transplants.

The primary outcome measures for this study were cold ischemia time and the incidence of DGF, defined as the need for dialysis in the first post-transplant week. Secondary outcome measures included rate of donor kidney biopsy, recipient wait time, transplant length of stay, post-transplant creatinine and creatinine clearance, incidence of biopsy-proven acute rejection, and actuarial graft and patient survival. Our study protocol was approved by the UCSF Committee on Human Research.


Induction therapy in the form of either IL-2 receptor blockade (daclizumab or basilixumab) or anti-thymocyte antibody was given at the discretion of the transplant surgeon to recipients with high immunologic risk, recipients with delayed or slow graft function, and/or recipients receiving steroid-sparing immunosuppresion. All patients received intravenous corticosteroids beginning in the operating room followed by tapering doses of oral prednisone. However, late during the study period, our center offered a steroid-avoidance protocol (induction with IL-2 receptor antibody with rapid 5-day corticosteroid taper) for low immunologic risk candidates. Mycophenolate mofetil was started at 2 g per day and adjusted for hematological abnormalities and/or clinical symptoms. A calcineurin inhibitor (either cyclosporine or tacrolimus) was added only when good renal function was present. Some recipients with poor initial renal function received sirolimus, typically as a bridge to, but occasionally in place, of calcineurin inhibitors.

Statistical analysis

ECD recipients (n = 20) and historical controls (n = 54) were compared using the Mann-Whitney test for continuous variables and Fisher's exact or chi-squared tests for categorical variables. Logistic regression was used to identify risk factors for DGF. Risk factors in the univariate model that were either significantly associated (p < 0.05) or trended toward association (p < 0.10) with DGF were included in the multivariate model. Kaplan-Meier actuarial time to first rejection and overall graft survival was calculated and compared using the log-rank test. Statistical significance throughout was defined as a p-value <0.05. All analyses were performed using SAS, version 9.0 software (SAS Institute, Cary, N.C.).


Dynamics of candidate response and the evaluation process

At the conclusion of the study period, a total of 356 wait-listed kidney transplant candidates who met ECD program criteria were contacted regarding the program. Three hundred twenty three (91%) of these candidates were sent letters and 33 (9%) were given information during their initial evaluation visit. Their responses and disposition are shown in Figure 1. The majority (231 patients; 65%) did not respond, 18 declined and 107 consented. Of consenting participants, 32 (30%) completed all testing required by our center's evaluation protocol and were approved to proceed with transplantation. The mean time from receipt of informed consent to approval for transplantation was 185 ± 103 days. Of the 32 eligible candidates, 23 underwent transplantation and nine remained on the waitlist by the completion of the study period. Twenty (87%) underwent ECD transplantation but three did not (one received a non-ECD kidney, one received a kidney from a donor after cardiac death, and one received dual kidneys). Candidates received ECD kidneys on average 275 ± 103 days after providing informed consent. Of the remaining 75 consenting candidates, 50 were still undergoing evaluation at the end of the study period while 25 were excluded because of death, withdrawal from program, diagnosis of malignancy, or medical unsuitability.

Figure 1.

Expanded criteria donor (ECD) candidate enrollment and outcome. The individual outcomes of eligible ECD candidates at the end of the study period are shown. Of 356 eligible candidates, 20 received ECD transplants over 1 year.

Donor, recipient and transplant characteristics

During the 1 year study period, ECD transplants constituted 11% (20 of 183) of all adult deceased-donor transplants performed at our center. By comparison, ECD transplants constituted 7% (54 of 730) in the preceding 5½ years. Donor characteristics for the two ECD transplant cohorts are compared in Table 2. Donors were similar in gender, ethnicity, weight, and body mass index, although the recent cohort of ECD donors was taller (172.9 ± 10.8 cm vs 167.8 ± 8.1 cm; p = 0.047) and older (62.8 ± 6.8 vs 59.2 ± 5.1 years; p = 0.052). The two cohorts were medically similar, with no differences in the history of hypertension, cause of death, and procurement creatinine. The mean relative risk for graft loss for the two donor cohorts was similar (2.06 vs 1.99; p = 0.282).

Table 2. Donor characteristics for recent and historical ECD transplants
 Recent ECD
recipients (n = 20)
Historical controls

  1. 1Mann-Whitney test for continuous variables, Fisher's exact test or chi-squared test for categorical variables.

  2. 2Mean ± standard deviation.

  3. 3Cockcroft-Gault creatinine clearance: (140 – age) × weight/72 × creatinine; ×0.85 if female.

  4. 4SRTR = Scientific Registry of Transplant Recipients.

Age (years)262.8 ± 6.859.2 ± 5.10.052
Male: N (%)12 (60)26 (48)0.44
Ethnicity: N (%) 0.73
 African-American3 (15)5 (9) 
 Asian/Pacific Islander1 (5)6 (11) 
 Caucasian12 (60)35 (65) 
 Hispanic3 (15)7 (13) 
 Other1 (5)1 (2) 
Weight (kilograms)278.3 ± 17.778.2 ± 19.20.80
Height (centimeters)2172.9 ± 10.8167.8 ± 8.10.047
Body mass index (kg/m2)226.3 ± 6.427.8 ± 6.80.33
History of hypertension: N (%)12 (60)32 (59)1.00
Cause of death: N (%) 0.53
 Cerebrovascular accident17 (85)41 (76) 
 Other3 (15)13 (24) 
Creatinine (mg/dL)21.0 ± 0.21.0 ± 0.40.37
Creatinine clearance (mL/min)2,383.4 ± 23.189.1 ± 27.30.43
SRTR relative risk of graft loss22.06 ± 0.261.99 ± 0.220.28

Recipient characteristics are compared in Table 3. Although recent ECD recipients were older than historical controls (67.1 ± 7.4 vs 59.5 ± 9.1 years; p = 0.001), the two cohorts were otherwise similar in gender, ethnicity, weight, height, body mass index, etiology of renal failure, mode of dialysis, peak panel reactive antibody, and history of prior transplant.

Table 3. Recipient characteristics for recent and historical ECD transplants
 Recent ECD
recipients (n = 20)
Historical controls
(n = 54)

  1. 1Mann-Whitney test for continuous variables, Fisher's exact test or chi-squared test for categorical variables.

  2. 2Mean ± standard deviation.

  3. 3CAPD = continuous ambulatory peritoneal dialysis; PRA = panel reactive antibody.

Age (years)267.1 ± 7.459.5 ± 9.10.001
Male: N (%)11 (55)27 (52)0.80
Ethnicity: N (%) 0.46
 African-American3 (15)14 (26) 
 Asian/Pacific Islander5 (25)10 (18) 
 Caucasian7 (35)21 (39) 
 Hispanic5 (25)6 (12) 
 Other0 (0)3 (5) 
Weight (kilograms)263.3 ± 8.972.3 ± 18.00.064
Height (centimeters)2165.2 ± 10.7168.3 ± 9.40.24
Body mass index (kg/m2)223.4 ± 2.625.3 ± 5.10.10
Etiology of renal failure: N (%)
 Diabetes8 (40)13 (24)0.25
 Other12 (60)41 (76) 
Dialysis mode: N (%) 0.53
 Hemodialysis13 (65)40 (74) 
 CAPD335 (25)12 (22) 
 None2 (10)2 (4) 
Peak PRA3 > 30%: N (%)0 (0)3 (6)0.56
Prior kidney transplant: N (%)0 (0)4 (7)0.57

Transplant characteristics are shown in Table 4. There was no difference in the degree of HLA matching between cohorts. The rate of donor kidney biopsy requested by any transplant center decreased from 85 to 24% (p < 0.001). At our center, mean cold ischemia time for ECD kidney transplants decreased from 16.4 ± 5.5 h to 7.4 ± 1.8 h (p < 0.001). By comparison, the mean cold ischemia time for non-ECD kidney transplants decreased from 17.6 ± 7.2 to 14.3 ± 6.8 h. There was a trend toward improvement in mean warm ischemia time (34.6 ± 10.4 min to 29.8 ± 6.8 min; p = 0.057).

Table 4. Transplant characteristics for recent and historical ECD transplants
 Recent ECD
recipients (n = 20)
Historical controls
(n = 54)

  1. 1Mann-Whitney test for continuous variables, Fisher's exact test or chi-squared test for categorical variables.

  2. 2Mean ± standard deviation.

  3. 3HLA = human leukocyte antigen.

Number of HLA mismatches2,34.5 ± 1.24.5 ± 1.30.92
Kidney biopsy: N/no. of donors (%)4/17 (24)35/41 (85)<0.001
Cold ischemia time (h)27.4 ± 1.816.4 ± 5.5<0.001
Warm ischemia time (min)229.8 ± 6.834.6 ± 10.40.057
Induction: N (%) 0.002
 IL2-R antibody14 (70)14 (26) 
 OKT®3 or Thymoglobulin®2 (10)20 (37) 
 None4 (20)20 (37) 
Steroid sparing regimen: N (%)5 (25)0 (0)0.002
Calcineurin inhibitor at discharge: N (%) 0.11
 Tacrolimus12 (60)23 (43) 
 Cyclosporine8 (40)21 (40) 
 None0 (0)9 (17) 
Sirolimus at discharge: N (%)1 (5)13 (24)0.094

Post-transplant immunosuppression evolved over the study period. A larger percentage of the recent cohort received induction immunosuppression (80% vs 63%, p = 0.002). Use of IL-2 receptor blockade increased from 26 to 70% while use of Thymoglobulin® or OKT®3 decreased from 37 to 10%. A steroid-sparing strategy, in which the dose was rapidly tapered to zero over 5 days, was used in 5 (25%) of the recent ECD cohort but none of the historical cohort (p = 0.002), as it was not an established immunosuppression protocol at our center during the early part of the study period. Although the use of tacrolimus and cyclosporine remained relatively stable between the two cohorts, there was a trend toward more patients in the historical cohort being discharged on sirolimus (24% vs 5%, p = 0.094), most often as a bridge to calcineurin inhibitors.

Recipient outcome

Recipient outcomes are shown in Table 5. Wait times for recent ECD recipients were 2.3 years less than those of historical controls (1.8 ± 0.9 vs 4.1 ± 1.4 years, p < 0.001). The incidence of DGF decreased from 43 to 15% (p = 0.031). There was a trend toward reduced length of hosptial stay, from 8.0 ± 4.8 days to 5.7 ± 3.7 days (p = 0.055). No significant differences were observed in serum creatinine at 3, 6 or 12 months or creatinine clearance at 12 months. Actuarial 1 year rejection-free survival, graft survival and patient survival were similar for the two cohorts (Table 5 and Figure 2).

Table 5. Recipient outcomes for recent and historical ECD transplants
 Recent ECD
recipients (n = 20)
Historical controls
(n = 54)

  1. 1Mann-Whitney test for continuous variables, Fisher's exact test or chi-squared test for categorical variables.

  2. 2Mean ± standard deviation.

  3. 3Cockcroft-Gault creatinine clearance = (140 – age) × weight/72 × creatinine; ×0.85 if female.

  4. 4Results given as actuarial survival percentage ± standard error.

Waitlist time (years)21.8 ± 0.94.1 ± 1.4<0.001
Delayed graft function: N (%)3 (15)23 (43)0.031
Length of stay (days) 225.7 ± 3.78.0 ± 4.80.055
 Immediate graft function4.5 ± 1.16.0 ± 2.40.024
 Delayed graft function12.3 ± 6.811.1 ± 5.80.73
Creatinine: 3 months (mg/dL)21.5 ± 0.41.9 ± 1.40.67
Creatinine: 6 months (mg/dL)21.5 ± 0.41.7 ± 0.70.34
Creatinine: 1 year (mg/dL)21.5 ± 0.41.6 ± 0.60.99
Creatinine clearance: 1 year (mL/min)2,344 ± 1251 ± 160.16
One year rejection-free survival4 (%)85 ± 873 ± 60.32
One year graft survival4 (%)85 ± 1085 ± 50.81
One year death-censored graft survival4 (%)100 ± 090 ± 100.22
One year patient survival4 (%)85 ± 1094 ± 40.37
Figure 2.

Kaplan-Meier curves of rejection-free survival and graft survival for recent ECD kidney recipients and historical controls. Death with a functioning graft was considered a graft loss. No statistically significant differences were noted between cohorts.

Risk factors for DGF

We hypothesized that the reduction in DGF observed after ECD allocation primarily resulted from the marked decrease in cold ischemia time rather than other factors. Logistic regression analyses were used to identify donor, recipient and transplant factors associated with DGF among all ECD transplants, both historical and current. Univariate analysis showed that six variables were significantly associated with DGF: donor age, recipient female gender, recipient height, recipient weight, HLA mismatch, and cold ischemia time (Table 6). Two additional variables, donor history of hypertension and increased recipient BMI, trended toward association with increased DGF. All variables with p < 0.10 were included in the multivariate model for DGF; recipient BMI was excluded since recipient height and weight were both included. Multivariate analysis showed that increased recipient height, increased HLA mismatching, and increased cold ischemia time were independently associated with increased DGF (Table 7). All of these associations were highly significant. Since recipient height and HLA mismatch did not differ between the historical and current ECD cohorts while cold ischemia time did (Tables 3 and 4), we suggest that the decrease in cold ischemia time likely explains the decrease in DGF incidence observed for the current ECD cohort.

Table 6. Univariate analysis for delayed graft function in 74 ECD recipients
CharacteristicOdds ratio95% CIp-value
  1. 1AAmerican = African American.

Donor age (per year)0.890.81–0.980.019
Donor female1.750.67–4.600.25
Donor ethnicity: AAmerican1 vs others2.000.34–11.720.18
Donor height (per 10 cm)0.960.91–1.020.17
Donor weight (per kg)1.010.99–1.040.36
Donor body mass index (per 1.0 kg/m2)1.050.98–1.130.15
Donor cause of death: stroke vs others0.880.28–2.760.82
Donor history of hypertension2.500.89–7.030.083
Donor creatinine (per 0.1 mg/dL)0.820.20–3.480.79
Donor creatinine >1.5 mg/dL0.920.16–5.380.92
Donor creatinine clearance (per mL/min)1.021.00–1.030.13
Recipient age (per year)0.990.94–1.040.66
Recipient female0.240.09–0.680.0075
Recipient ethnicity: AAmerican1 vs others1.370.41–4.640.42
Recipient height (per 10 cm)1.131.06–1.20<0.001
Recipient weight (per kg)1.061.03–1.110.0013
Recipient body mass index (per 1.0 kg/m2)1.171.00–1.250.052
Donor-to-recipient body mass index ratio0.950.26–3.470.94
Recipient disease: diabetes vs all others1.200.42–3.410.74
Recipient dialysis: hemodialysis vs other0.380.11–1.290.12
HLA mismatch: 5 or 6 mismatches vs 0–44.201.36–12.970.013
Cold ischemia time (per hour)1.151.05–1.250.0029
Warm ischemia time (per minute)1.030.98–1.080.29
Received sirolimus prior to post-txp dialysis4.180.71–24.600.11
Transplant year0.880.67–1.150.33
Table 7. Multivariate analysis for delayed graft function in 74 ECD kidney recipients
CharacteristicOdds ratio95% CIp-Value
Donor age (per year)0.850.69–1.0460.13
Donor history of hypertension5.010.54–46.870.16
Recipient female gender0.460.077–2.720.39
Recipient height (per 10 cm)1.211.05–1.380.0080
Recipient weight (per kilogram)1.000.95–1.050.96
HLA mismatch: 5–6 mismatches vs 0–420.462.74–152.940.0033
Cold ischemia time (per hour)1.241.07–1.430.0036


In response to the new UNOS ECD allocation policy that came into effect on October 31, 2002, our transplant center launched an ECD program aiming to identify, obtain consent from, and evaluate appropriate wait-listed candidates for ECD kidney transplantation. During the first calendar year, 20 candidates underwent ECD kidney transplantation. Compared to a historical cohort of retrospectively identified transplants that met the ECD definition, the new cohort had a substantial reduction in cold ischemia time, which correlated with a significant reduction in the incidence of DGF. Therefore, at least for our transplant center, ECD kidney allocation policy achieved one of its intended goals, facilitating organ placement to reduce cold ischemia time.

We believe that careful candidate selection is important for ECD kidney transplantation. We used recommendations from a national consensus conference to develop inclusion and exclusion criteria for our ECD program (7). Although nearly all wait-listed dialysis patients derive a survival benefit from an ECD kidney (4), we chose to offer ECD kidneys preferentially to candidates who faced substantial morbidity and/or mortality on dialysis, such as diabetic and/or older candidates. Diabetics with end-stage renal disease are the subgroup known to face the highest annual mortality on dialysis and to derive the greatest benefit from kidney transplantation (1,4). As for older candidates, precedent has been set by the 'old-for-old' donor-recipient matching policy in Europe, which has successfully expanded the donor pool without compromising recipient outcome (11,12). Moreover, placing older kidneys with less nephron mass into older recipients who require fewer nephrons achieves some degree of functional matching, a principle we further served by setting limits on recipient body weight. Donor-to-recipient size disparities can adversely affect graft survival (13–15) and ECD kidneys can be thought of as comparable to kidneys from small donors, since they share the quality of diminished functional reserve (16). Furthermore, greater emphasis on matching nephron mass as a strategy to optimize outcomes of ECD kidney transplantation has been recently reported (17).

In addition to candidate disease etiology, age and size, we also considered the candidates' accumulated wait time, which reflects their access to non-ECD kidney transplantation. Since ECD kidney transplantation offers longer survival compared to dialysis but shorter survival compared to non-ECD kidney transplantation, we designed our program to shorten wait time by approximately 2 years. Admittedly, our local variance to allocate kidneys by waiting time alone increases the predictability of the time to non-ECD kidney transplantation. However, since the majority of kidneys are transplanted into unsensitized patients with the longest waiting time, all centers should be able to estimate a candidate's time to non-ECD transplantation (7,18). Schnitzler et al. used Markov modeling to analyze the decision to accept or refuse an ECD kidney offer (19). The benefit of an ECD kidney transplant outweighed remaining on the waitlist only when the anticipated remaining wait time was ≥3.2 years. However, for candidates >60 years who have higher mortality on dialysis, the break-even point decreased to 0.9 years. In contrast, for sensitized recipients for whom ECD kidneys have disproportionately poorer graft survival than non-ECD kidneys, the break-even point increased to 5.2 years. These considerations, along with the additional time often required for crossmatching (which can increase cold ischemia time), motivated us to exclude sensitized candidates from ECD kidney transplantation.

All transplant candidates who met our inclusion and exclusion criteria were given materials explaining the ECD program and a consent form, either by mail or in person. The vast majority (91%) of these candidates had already been listed for transplant and received information through the mail, while the remainder were given information during their initial evaluation visit and asked to return the signed consent by mail to indicate willingness to participate. Our high no-response rate (65%) suggests that mass mailing is not an effective communication strategy. Of the 107 (30%) of candidates who consented and began to be evaluated, only half (57/107) reached a final disposition by the end of the study. This was primarily the result of a high rate of co-morbid disease in ECD candidates, who frequently required extensive pre-transplant evaluation for cardiac and vascular disease (20). Typically, 6 months were required to complete pre-transplant evaluation. Since candidates spent, on average, 3 additional months awaiting a ECD kidney offer, the overall time to from consent to transplantation averaged 9 months.

Our first ECD kidney transplant occurred on July 28, 2003, nearly 9 months after national allocation for ECDs went into effect and 7 months after local ECD allocation began. The 20 candidates who underwent ECD transplantation during the ensuing calendar year waited, on average, 2.3 years less than the 54 candidates in the historical cohort (p < 0.001). This reduction in recipient wait time was a principal benefit of the ECD program (2,3). If demand and competition for ECD kidneys escalates in the future, then ECD waiting times might lengthen and this benefit could erode. We believe, however, that this benefit can and should be preserved. If access to a non-ECD kidney is predictable, then ECD kidney candidacy should be periodically reconsidered. Candidates who are close to receiving a non-ECD kidney can be removed from the ECD list. Removal of such candidates is currently a provision of our ECD program.

Comparison of outcomes for the recent versus historical ECD cohorts showed a marked reduction in cold ischemia time from 16.4 ± 5.5 to 7.4 ± 1.8 h (p < 0.001). Although there was also a decrease in cold ischemia time for non-ECD transplants (17.6 ± 7.2 to 14.3 ± 6.8 h) at our center, the magnitude of the change was much less. Several factors likely contributed to this change. First, there was a substantial drop in the rate of donor kidney biopsy from 85 to 24% (p < 0.001). Considering the conflicting literature regarding the usefulness of donor biopsy (21–23), we opted to rely on the donor's relative risk rather than biopsy to assess kidney quality. Kidney biopsy can delay final organ placement and thereby increase cold ischemia time. Second, we excluded sensitized candidates who require more extensive crossmatching, frequently delaying kidney placement. Third, we accelerated the complete evaluation of pre-selected and pre-consented candidates to ensure that kidneys were only offered to candidates willing and ready to proceed with ECD transplantation. Finally, the recent attention on ECD kidneys, which emphasized the importance of minimizing cold ischemia time, may well have resulted in practice changes that expedited actual transplantation after a kidney had been placed with a recipient.

Concomitant with the decrease in cold ischemia time was a decrease in DGF. Our regression analyses confirmed that the shorter cold ischemia time was independently associated with less DGF. Cold ischemia time clearly differed between the historical and the current ECD cohorts (Table 4). Therefore, we suggest that the decrease in cold ischemia time was predominantly responsible for the decrease in DGF for the current cohort. Other factors that protected against DGF for all ECD transplants included improved HLA matching and several surrogates for decreased nephron demand, of which only recipient height retained significance in the multivariate model. These factors, however, did not differ significantly between the current and historical cohorts and therefore, would be unlikely to explain the markedly different incidence of DGF (Table 3 and 4). The impact of poor HLA matching (five or six mismatches) was extremely and, perhaps surprisingly, potent. It is possible that the increased susceptibility of older donor kidneys to ischemia/reperfusion injury may magnify the impact of HLA mismatching (24). This association should be investigated further, particularly since decreased emphasis on HLA matching is a cornerstone of the current ECD allocation policy. It was not surprising that no donor factors were significant predictors of DGF after ECD kidney transplantation. As for the effect of immunosuppression, it is believed that sirolimus may affect the incidence and/or duration of DGF (25,26). For our historical cohort, sirolimus was often given as a strategy to avoid calcineurin inhibitors to patients with poor early graft function. Typically, sirolimus was initiated within 2 days of transplantation, when it was clear how well the allograft was functioning. This accounts for the near association that we observed between sirolimus and DGF in this study. We abandoned this immunosuppression strategy in 2002, after finding that sirolimus prolonged recovery from DGF. Finally, the trend toward decreased length of stay for the recent ECD cohort (2.3 days shorter than for the historical cohort) can be attributed, in part, to a general decrease in the length of stay for non-DGF recipients and, in part, to the fewer numbers of DGF recipients.

In conclusion, our center has had a favorable experience with ECD allocation. The ECD designation can be considered a good description of kidney quality, which can essentially eliminate biopsy-related delays in organ placement and transplantation. We found that ECD kidney allocation to pre-selected, pre-consented, and medically-ready transplant candidates can shorten cold ischemia times and decrease DGF. Whether avoidance of DGF will improve graft function and survival remains to be seen.