Center-Level Patterns of Indicated Willingness to and Actual Acceptance of Marginal Kidneys

Authors


Corresponding author: Dorry L. Segev, dorry@jhmi.edu

Abstract

UNetSM, the UNOS data collection and electronic organ allocation system, allows centers to specify organ offer acceptance criteria for patients on their kidney waiting list. We hypothesized that the system might not be fully utilized and that the criteria specified by most transplant centers would be much broader than the characteristics of organs actually transplanted by those centers. We analyzed the distribution of criteria values among waitlist patients (N = 304 385) between January 2000 and February 2009, mean criteria values among listed candidates on February 19, 2009 and differences between a center's specified criteria and the organs it accepted for transplant between July 2005 and April 2009. We found wide variation in use of criteria variables, with some variables mostly or entirely unused. Most centers specified very broad criteria, with little within-center variation by patient. An offer of a kidney with parameters more extreme than the maximum actually transplanted at that center was designated a ‘surplus offer’ and indicated a potentially avoidable delay in distribution. We found 7373 surplus offers (7.1% of all offers), concentrated among a small number of centers. The organ acceptance criteria system is currently underutilized, leading to possibly avoidable inefficiencies in organ distribution.

Abbreviations: 
BMI

body mass index

CIT

cold ischemia time

DCD

donation after cardiac death

DSA

donation service area

ECD

expanded criteria donor

HBV

hepatitis B virus

HCV

hepatitis C virus

HTLV

human T-lymphotropic virus

IQR

interquartile range

MAC

minimum acceptance criteria

UNet

United Network for Organ Sharing's data collection and electronic organ allocation system

UNOS

United Network for Organ Sharing

Background

When a deceased donor kidney becomes available, it must be allocated and distributed as quickly as possible in order to minimize cold ischemia time (CIT) and the risk of discard (1). This principle of efficiency must coexist with that of equity, where patients at different transplant centers should have an equal chance of being offered an organ (2). However, each offer refused by a given transplant center in the pursuit of equity potentially incurs a cost in efficiency, not only in administrative costs to the center which was offered the organ, but also in CIT which might worsen the outcome of the patient who ultimately receives the organ. Rates of end-stage renal disease, live-donor kidney transplantation and deceased donor organ donation all vary across geographic areas, leading to varying waitlist time across donation service areas (DSAs) (3–6). Moreover, transplant centers may differ in their willingness to accept organs of marginal perceived quality (1). Within a given transplant center, patients differ in appropriateness for kidneys of varying quality (7,8).

The UNOS data collection and electronic organ allocation system1 (UNetSM) allows centers to specify, in advance, donor acceptance criteria for organs they are willing to accept for transplantation. A center can specify default criteria, which will apply to all registrants at that center; however, it may override these defaults with different values for an individual registrant. Organs from donors who do not meet the acceptance criteria will not be offered to the patient/center, assuming the characteristic has been entered into DonorNet before the match run (entry of some donor characteristics is not required before the match run). Therefore, in principle, such organs will be offered more quickly to a center willing to accept them. These criteria became particularly important with the introduction of DonorNet 2007®2 (9). For kidneys, the possible acceptance criteria include: maximum warm ischemia time; maximum CIT; maximum percent sclerosis; minimum and maximum donor weight; peak and terminal serum creatinine; minimum and maximum donor age; donor diabetes; donor hypertension; donation after cardiac death (DCD); and positive tests for hepatitic C virus (HCV), hepatitis B virus (HBV) core antigen, and human T-lymphocytic virus (HTLV). Criteria for donor age, serum creatinine, sclerosis and DCD can be specified separately for kidneys that become available outside of a center's DSA.

Such a system is only effective when centers specify preset criteria in accordance with actual organ acceptance practices. For example, a center may list a patient as willing to accept DCD organs, but reject all DCD organs offered to that patient. The effect on the patient is the same, but the administrative burden on the center and UNOS is higher, and ultimately CIT of DCD organs may increase because of the delays incurred by rejected offers. We hypothesized that centers would specify criteria that were broader than the range of organs that a center was actually willing to accept. This might occur for three reasons. First, there may be a fear that regulatory agencies (or lawyers representing patients on the waiting list) might criticize centers with stringent acceptance criteria. Second, providers may not be conscious of the framework by which they determine organ acceptability. Third, currently available criteria may be inadequate to appropriately reflect the decision tree that providers use when evaluating organs. We further hypothesized that, although centers are allowed to specify individual criteria for each registrant, in practice most centers would use the same criteria for each adult registrant.

To test our hypotheses, and to describe a component of kidney distribution that, to our knowledge, has never before been rigorously analyzed, we conducted a 9-year retrospective study of national patterns of donor acceptance criteria use. Our goals were (1) to describe the use of donor acceptance criteria over time, (2) to analyze within-center heterogeneity of prespecified criteria and (3) to compare the prespecified criteria to actual patterns of organ acceptance.

Methods

Study population

We analyzed donor acceptance criteria in a longitudinal data set collected prospectively by the Organ Procurement and Transplantation Network (OPTN) between January 1, 2000 and February 19, 2009. Criteria for a registrant exist from the time of listing to the time of removal from the waiting list, and may be changed by the center at any time. To study patterns over time, we created summary variables for each center-month of: mean value of each criterion, proportion of missing values for each criterion, mode of each criterion (the most common value specified by that center during that month) and proportion of all values (among adult registrants at that center during that month) which were equal to the mode value for that center month. Person–time accrued before the registrant reached age 18 was analyzed separately. We also analyzed a cross-sectional data set of donor acceptance criteria for patients on the waiting list on February 19, 2009.

To examine the relationship between criteria specified by a center and a center's behavior in accepting organ offers, we analyzed organ offers made between July 1, 2005 and April 18, 2009, also prospectively collected by UNOS. Offers from pediatric donors, offers to pediatric recipients, organs allocated in an expedited manner because of zero-HLA-mismatch, and offers determined to be administrative bypasses were excluded.

Handling of missing data was informed by the way missing values are used for organ distribution. UNetSM interprets a missing criterion value as accepting all organs. Therefore, a missing minimum value criterion was set to zero, a missing maximum value criterion was set to a high sentinel value (e.g. missing maximum donor age would be set to 99 years), and a missing binary criterion (e.g. accept DCD kidneys) was set to one (i.e. ‘accept any’). These sentinel values were not used in calculating mean criteria values, but they were used to compare donor acceptance criteria with actual acceptance practices.

Characteristics of the donor acceptance criteria data set

Because many criteria values were blank, particularly in the early years of our study period, we first graphed the proportion of records for each criterion which were set to any value at all. We grouped these into three categories: heavily used (with a value specified for >99% of registrants nationally), gradual uptake over time (nonzero usage from the beginning of our study period, with increasing usage over time) and recently added (criteria added to the system by UNOS after January 2000). We limited our analysis of mean-value trends over time to heavily used criteria (as defined above).

To describe donor criteria for patients in the cross-sectional data set, we computed mean and standard deviation for each continuous criterion. We used linear regression to test the hypothesis that different criteria might be used for pediatric registrants than for adults, with a clustered sandwich estimator for the standard error to account for possible intracenter correlation. Thus we obtained a ‘pediatric coefficient’ for each criterion, representing an estimate of the difference between the mean values for pediatric versus adult registrants. For binary criteria, we computed the proportion of adult and pediatric patients with a permissive value (‘1’ or missing), and calculated the relative rate of permissive values for pediatric versus adult registrants using Poisson regression with inflated variance to account for center-level clustering. The regression models gave a ‘pediatric relative rate’ representing the ratio of the proportion of pediatric registrants for whom a category of organ was listed as acceptable, to that proportion in adults.

Homogeneity of criteria within a center

We examined the possible tendency of a center to use the same criteria values for most or all registrants by calculating the percentage of values for each criterion in the cross-sectional data set that were equal to the mode for that center. We also counted, for each criterion, the number of centers that had the same value for all patients.

Donor criteria and organ acceptance

To compare a center's practices in specifying donor criteria to its behavior in accepting organ offers, we produced scatter plots of each center's mode value for maximum permissible donor age and terminal serum creatinine on February 19, 2009 against the highest value accepted by that center between July 2005 and April 2009. If a center's mode value for a criterion exceeded the highest value accepted by that center over the 4-year period (e.g. the mode for the terminal creatinine criterion was 5.0 mg/dL, but the center accepted no kidneys with creatinine above 1.8), then physicians at that center might have been making decisions based on a threshold lower than the one they specified for that criterion. We described the difference between the mode criterion value and the maximum value of transplanted kidneys at that center as the center's overestimate of its criterion. Thus in the example above, the center's overestimate of its terminal creatinine criterion would have been 3.2 mg/dL. The ‘overestimate’ could be negative if a center's criteria tighten over time (for example, if the center was more aggressive in accepting high-creatinine kidneys in the past) or if a different decision-making strategy was used for different patients (e.g. if the center considered kidneys with terminal creatinine above 5.0 mg/dL acceptable for only a few of its patients). This is a conservative assessment: even if all other kidneys accepted at a center had creatinine of 1.0, just one kidney with creatinine of 5.0 would have excluded that center from an ‘overestimate’. We produced histograms of each center's overestimate of each criterion.

Surplus offers

We used the term surplus offer to describe an offer of a kidney with characteristics outside the bounds of characteristics of kidneys actually transplanted by a center. In the example above, any offer from a donor with creatinine greater than 1.8 mg/dL would be a surplus offer. Similarly, a surplus offer based on DCD was defined as any DCD kidney offered to a center which accepted no DCD kidneys from July 2005 to April 2009, but which designated DCD kidneys as acceptable for a majority of its patients on the waiting list on February 19, 2009. Again, this is conservative: if a center transplanted even one DCD kidney, it will have zero ‘surplus’ offers for DCD. By definition, a refused offer cannot be ‘surplus’ if the kidney was accepted for transplantation into another patient at the same center. We calculated surplus offers based on the criteria of maximum permissible donor age, terminal creatinine, hypertension, diabetes and DCD. The offer of the same kidney to multiple patients at a single center constituted a single offer. For each criterion, we calculated the median and interquartile range (IQR) of surplus offers to each center as well as the total number of surplus offers to all centers. Because some surplus offers may have come from donors whose kidneys would not be considered by any transplant center, we calculated the percentage of all surplus offers that came from donors from whom at least one kidney was accepted for transplant.

To explore the distribution of surplus offers across transplant centers, we created a histogram of the total number of surplus offers made to each center. We also created a histogram of the percentage of all offers that were surplus, among centers that received at least 100 offers (for statistical stability).

Statistical analysis

All calculated p-values were two sided. Statistical analysis was performed using Stata/MP 11.0 for Linux (StataCorp LP, College Station, TX, USA). Confidence intervals are reported using the method of Louis and Zeger (10).

Results

Study population

The longitudinal kidney acceptance criteria data set covered 7 456 841 person months for 304 385 registrants, including 297 012 adult and 7859 pediatric patients. Criteria for adult registrants were grouped into 26 686 center months of center-level records. The cross-sectional data set contained criteria for 82 314 adult and 975 pediatric patients on the kidney waiting list on February 19, 2009.

Characteristics of the donor acceptance criteria data set

Three criteria were defined for >99.9% of records continuously over our study period: maximum and minimum donor age and acceptance of HCV+ organs (Figure 1A). Maximum acceptable warm ischemia time gradually increased in usage (percentage of records with a defined nonmissing value) from about 20% of records in 2000 to about 80% of records in 2009 (Figure 1B). Values for CIT, sclerosis and serum creatinine criteria were initially defined for about 90% of records, but decreased in usage until 2003, at which point usage steadily increased, nearing 100% by February 2009. All of these variables showed a spike in usage around late 2007, likely corresponding to implementation of DonorNet 2007®. The DCD criterion, introduced in September 2007, saw rapid uptake, with 80% usage by early 2008. Usage of the diabetes, hypertension, HTLV and body mass index (BMI) criteria, introduced in July 2008, climbed from zero to 10–28% (Figure 1C). All heavily used criteria tightened over time, notably acceptance of HCV+ organs which were within organ offer acceptability range for more than 10% of patients in 2000, but were considered for less than 3% of patients in 2009 (Figure 2).

Figure 1.

Patterns of usage of donor criteria over time among adult waitlist registrants (N = 7 349 674 center months for 297 012 registrants). Each graph shows, for a given donor criterion, the proportion of patients for whom the criterion was defined to be any value (nonmissing). Criteria are sorted into three categories, based on the proportion of patients over time for whom a value was assigned.

Figure 2.

Change of mean values of max and min criteria (upper panel) and HCV+ criterion (lower panel) over time for adult waitlist registrants. All criteria have tightened over time. Acceptance of HCV+ organs has dropped from above 10% of patients to less than 3% of patients.

Means and standard deviations of continuous donor acceptance criteria (for which a center specifies a maximum or minimum acceptable value) on February 2009 are listed in Table 1. For criteria involving a maximum acceptable value (e.g. CIT, with an offer only being made if CIT at time of offer is less than the value set as a criterion), the criterion is referred to as a ‘ceiling’. Thus the ‘mean CIT ceiling’ refers to the average, over all registrants on the national waiting list, of the maximum CIT at which the registrant's center specified that an organ might be offered to that registrant. A criterion involving a minimum acceptable value (e.g. donor age) is referred to as a ‘floor’.

Table 1.  Summary of continuous donor criteria, among patients on the deceased donor kidney waiting list on February 19, 2009
CriterionN (%) definedAdult mean (SD)Pediatric coefficient
AdultPediatric
  1. ‘N defined’ refers to the number of patients for whom the criterion was specified. If a criterion was unspecified, then kidneys would be offered to the patient regardless of the donor's value of that criterion. ‘Adult mean’ refers to the mean value among adult registrants for whom the value was specified. ‘Pediatric coefficient’ refers to the difference between the mean value for adult and pediatric registrants; negative values indicate that the mean value was lower for pediatric patients.

  2. 1Denotes a criterion applied only to offers that are regional/national shares.

  3. *p < 0.05.

  4. **p < 0.01.

  5. ***p < 0.001.

Warm ischemia time (mins)70 853 (86.1%)906 (92.9%)45.33 (19.50)−16.64−10.19−3.73 **
CIT (hrs)80 633 (98.0%)970 (99.5%)35.17 (12.47)−7.64−4.7−1.75** 
Min donor age82 314 (100.0%)975 (100.0%)1.49 (2.14)  0.762.353.94** 
Min age (import)182 314 (100.0%)975 (100.0%)1.78 (2.35)  0.442.073.69* 
Max donor age82 314 (100.0%)975 (100.0%)76.50 (9.87)−23.19−19.68−16.16***
Max age (import)82 314 (100.0%)975 (100.0%)74.32 (10.26)−21.23−17.72−14.21***
Sclerosis (<10 glomeruli)80 503 (97.8%)962 (98.7%)27.60 (18.45) −11.58−6.44−1.31*  
Sclerosis (≥10 glomeruli)80 602 (97.9%)962 (98.7%)25.76 (15.55)−9.77−6.35−2.92***
Sclerosis (<10 glomeruli) import180 503 (97.8%)962 (98.7%)27.22 (18.71) −11.33−6.19−1.05*  
Sclerosis (≥10 glomeruli) import180 602 (97.9%)962 (98.7%)25.30 (15.07)−9.96−6.49−3.01***
Peak serum creatinine80 590 (97.9%)968 (99.3%)5.30 (2.48)−2.06−1.47−0.88***
Final serum creatinine80 591 (97.9%)968 (99.3%)4.23 (2.47)−2.08−1.52−0.97***
Peak creatinine (import)180 590 (97.9%)968 (99.3%)5.02 (2.53)−1.88−1.27−0.67***
Final creatinine (import)180 591 (97.9%)968 (99.3%)3.98 (2.55)−1.93−1.36−0.79***
BMI (kg/m2)9000 (10.9%)137 (14.1%)53.84 (16.02) −20.29−11.44−2.58*   

In general the specified criteria were quite permissive. For example, for adults, the mean value of the CIT ceiling criterion is 35.2 h; mean BMI ceiling was 53.9 kg/m2; mean terminal serum creatinine ceiling was 4.2 mg/dL for local organs and 4.0 mg/dL for imports. All criteria were statistically significantly stricter for pediatric registrants; for example, the donor age ceiling is 16.219.723.2 years lower for pediatric registrants than for adult registrants. Nevertheless, even for pediatric registrants, the criteria were quite broad, with mean age ceiling of 56.6 years, mean terminal serum creatinine ceiling of 2.7 mg/dL, and mean BMI ceiling of 42.4 kg/m2. Note that many of the criteria listed in Table 1 are effectively even more permissive than might be gathered from examining means. As specified above, means were only calculated from among those registrations where the criterion was specified. Not specifying the criterion is the most permissive setting, so both the proportion where the criterion was specified and the mean need to be considered for an accurate understanding of the use of that criterion.

A summary of binary donor criteria (for which a center specifies whether or not a given category of organs is acceptable) appears in Table 2. Kidneys from diabetic or hypertensive donors, as well as locally available DCD kidneys, were listed as potentially acceptable to over 95% of adult registrants, and kidneys from HTLV+ patients were potentially acceptable to 73%. Again, all criteria were stricter for pediatric registrants than for adults. For example, children were 47% less likely than adults to be listed as considering DCD organs (pediatric relative rate 0.420.530.67), and 17% less likely to be listed as considering a kidney from a diabetic donor (0.770.830.89). The increased strictness for pediatric registrants was statistically significant for all criteria except for HCV+ (0.280.551.06) and HTLV+ (0.780.891.02).

Table 2.  Summary of binary donor criteria, among patients on the deceased donor kidney waiting list on February 19, 2009
CriterionPatients accepting an organ with the criterionPediatric relative rate
Adult (N = 82 314)Pediatric (N = 975)
  1. ‘Patients accepting an organ with the criterion’ includes patients with a missing value for this criterion, who would be offered an organ with the criterion by default. ‘Pediatric relative rate’ refers to the ratio of the proportion of pediatric registrants listed as accepting offers of organs with the criterion, to the proportion of adults listed as accepting such offers. In general, criteria were stricter for pediatric registrants, although organs with any binary criterion other than HCV+ or HBV core+ would be offered to most patients.

  2. 1Denotes a criterion applied only to offers that are regional/national shares.

  3. *p < 0.05.

  4. ***p < 0.001.

DCD79 719 (96.8%)498 (51.1%)−0.420.53−0.67***
DCD (import)165 602 (79.7%)371 (38.1%)−0.360.48−0.63***
Diabetes81 618 (99.2%)801 (82.2%)−0.770.83−0.89***
Hypertension81 872 (99.5%)856 (87.8%)−0.830.88−0.93***
HCV+ 2006 (2.4%)13 (1.3%)−0.280.55−1.06   
HBV core+43 492 (52.8%)358 (36.7%)−0.510.69−0.95*  
HTLV+60 339 (73.3%)636 (65.2%)−0.780.89−1.02   

Homogeneity of criteria within a center

As of February 19, 2009, over 90% of criteria values were equal to the mode value for that criterion in a given center (Table 3). For most criteria, over half of centers use the same value for all adult patients. The HCV+ criterion had one of the highest percentages of values equal to the center mode (98.0%), but one of the lowest percentages of centers with the same value for everyone (42.4%), likely indicating that this criterion is being used to match HCV+ donors to HCV+ recipients. Graphs of the proportion of values equal to the center mode in a given month suggest that the practice of using the same value for most patients was relatively constant over time (Figure 3).

Table 3.  Homogeneity of criteria within a center, based on organ-acceptance criteria specified in the deceased donor kidney waiting list on February 19, 2009
CriterionPercentage of values equal to center mode (%)Number of centers with the same value for all patients
  1. For each criterion, the most common value in each center (including missing values) is determined; column 2 contains the proportion of all registrants with the criterion set to that value. For all criteria, at least 90% of patients have the mode value, indicating that centers have a default value for donor criteria that they use for most patients.

  2. 1Denotes a criterion applied only to offers which are regional/national shares.

Warm ischemia time (mins)94.4161 (64.4%)
CIT (h)95.0159 (63.6%)
Min donor age (years)96.1137 (54.8%)
Min age (import)195.5139 (55.6%)
Max donor age92.1111 (44.4%)
Max age (import)192.1109 (43.6%)
Sclerosis (<10 glomeruli)93.0164 (65.6%)
Sclerosis (≥10 glomeruli)93.2168 (67.2%)
Sclerosis (<10 glomeruli) import192.8161 (64.4%)
Sclerosis (≥10 glomeruli) import194.5166 (66.4%)
Peak serum creatinine (mg/dL)90.1149 (59.6%)
Final serum creatinine92.6139 (55.6%)
Peak creatinine (import)191.1143 (57.2%)
Final creatinine (import)191.0140 (56.0%)
BMI (kg/m2)97.3193 (77.2%)
DCD97.8134 (53.6%)
DCD (import)195.0120 (48.0%)
Diabetes99.2170 (68.0%)
Hypertension99.5182 (72.8%)
HCV+98.0106 (42.4%)
HBV core+90.5114 (45.6%)
HTLV+92.5109 (43.6%)
Figure 3.

Change over time in proportion of donor criteria variable values which are equal to the mode value for a given center at a given time, among adult waitlist registrants. One-month downward spikes indicate a changing of all values for all patients within a month. The downward spike in late 2007 corresponds to implementation of DonorNet 2007. The second image is a detail of the last few years.

Donor criteria and organ acceptance

Unsurprisingly, the age of the oldest donor kidney accepted by a center since July 2005 was in most cases less than that center's mode age ceiling criterion in February 2009 (Figure 4). However, for many centers the difference between the age criterion and the oldest transplanted kidney was many years or even decades. A few centers transplanted organs with terminal serum creatinine exceeding their mode creatinine ceiling in 2009; exploration of transplants of kidneys with extremely high creatinine (>10 mg/dL) suggests that this criterion was not rigorously applied. Nevertheless, most centers’ serum creatinine ceiling values were considerably higher than the highest terminal serum creatinine value actually transplanted.

Figure 4.

The scatter plots on the left compare the mode value of each center as of February 2009 for donor age (A) and serum creatinine (B) to the maximum values for organs actually transplanted in that center between July 2005 and April 2009 (N = 242 centers). The histograms on the right show ‘overestimation’—the mode criterion, minus the actual maximum value in a transplanted organ.

Over 80% of centers transplanted at least one DCD kidney between July 2005 and April 2009; even more transplanted at least one hypertensive and one diabetic kidney over the same time period. Therefore, the binary criteria generated relatively few surplus offers, the most being 284 surplus offers across all centers for diabetes (Table 4). The continuous criteria generated many more surplus offers, with 3580 surplus offers for serum creatinine and 3570 for donor age. The total number of surplus offers was 7373, constituting 7.1% of 103 787 offers. This number is smaller than the sum of surplus offers defined by each criterion because an offer could be defined as surplus by more than one criterion (e.g. both serum creatinine and donor age exceed the maximum value accepted by a center). Of the 7373 offers, 5914 (80.2%) were offers either of a kidney which eventually got transplanted, or of a kidney from a donor whose other kidney was eventually transplanted.

Table 4.  The distribution of surplus offers for various donor criteria
 Center criteria acceptsCenter accepted at least oneMedian (IQR) surplus offersTotal surplus offers
  1. A ‘surplus offer’ is defined as an organ offer that exceeds the maximum value accepted by the center in the past 4 years. For hypertension, DCD and diabetes: ‘center criteria accepts’ shows, for binary criteria, the number of centers which were willing to consider such an organ on February 19, 2009; ‘center accepted at least one’ shows the number of centers which actually accepted at least one such organ from July 2005 to April 2009. Surplus offers existed for a binary criterion only if a transplant center accepted no offers of organs meeting that criterion.

Max donor age8.5 (2–20)3570 
Serum creatinine 9 (2–22)3580 
% glomerulosclerosis0 (0–2)538
Hypertension240 (99.17%)211 (87.19%)0 (0–0)261
DCD217 (89.67%)196 (80.99%)0 (0–0)258
Diabetes235 (97.11%)202 (83.47%)0 (0–0)284

The median number of surplus offers per center was 21, but the distribution of surplus offers skewed to the right (Figure 5A) and the mean was 30.5. Of 242 centers, 10 (4.1%) received more than 100 surplus offers, with one center receiving 189 surplus offers, or about one per week over the study period. By definition, all offers to centers which accepted no deceased donor kidneys for transplant were surplus; however, these were not a major contributor to the number of surplus offers, as no such center received more than 34 total offers. Among the 204 centers that received at least 100 offers over the study period, the median proportion of offers that were surplus was 5.7%. This distribution also skewed to the right (Figure 5B). The 75th percentile was 12.1%, and four centers that received more than 100 total offers received at least 40% surplus offers.

Figure 5.

Histograms of number of ‘surplus’ offers per transplant center (A) and percentage of all offers which were surplus (among the 204 centers that received at least 100 offers) (B). A surplus offer is defined as one that falls outside of criteria based on extreme values for organs that a center actually accepted. The median number of surplus offers was 21 surplus offers per center, but there was sharp skew to the right. The percentage of all offers received which were surplus also skewed to the right.

Discussion

We have shown that the donor criteria system is only partly utilized by transplant centers. Most centers use the same criteria values for almost all of their adult registrants. Although age criteria have narrowed over time, many centers still considerably overestimate their own willingness to accept organs from donors with advanced age or high serum creatinine.

The goal of the donor criteria system is to increase efficiency of organ allocation and distribution by not offering organs of perceived marginal quality to centers that do not want them. We have shown that centers set criteria broader than the range of organs they actually accept, resulting in thousands of surplus offers. Most of these surplus offers were of kidneys from a donor from whom at least one kidney was eventually transplanted, suggesting that reducing surplus offers would lead to real decreases in CIT. Because a handful of centers account for a sizeable proportion of all surplus offers, the number of surplus offers could be significantly reduced, and system-wide efficiency improved, by encouraging those few centers to tighten their acceptance criteria to reflect their actual decision making heuristics. It would seem reasonable for UNOS to inform centers if their specified criteria do not match observed acceptance practices.

We have also shown that centers use a default value instead of specifying patient-level criteria for most registrants. Either centers do not consider that some organs are beneficial for some patients but not for others, or they make most patient-level decisions at the time of the organ offer, thus incurring the administrative costs and CIT delays which the donor acceptance criteria system was designed to avoid. Setting patient-level criteria in advance, rather than at the time of offer, would save time in the long run by reducing unwanted offers.

One likely drawback of the current system is that criteria are defined separately. A center may be reluctant to rule out all kidneys from hypertensive donors, or from older donors, for patients who are in desperate need of a transplant. However, undesirable traits of organ donors are likely additive. With the exception of ECD listing (which did not form part of our analysis because it is specified through a separate system) (8), there is no way for clinicians to specify combinations of criteria, which would constitute an unacceptable organ. Incorporation of multifactorial metrics or decision trees into organ allocation may fill this gap. Once a better system is in place for specifying organ acceptance criteria that is better aligned with a center's heuristics for clinical decision-making, it might then be reasonable for UNOS to more rigorously enforce the use of these criteria.

Our analysis of the UNetSM acceptance criteria has not taken into account other mechanisms through which clinicians can limit offers of unwanted organs to registrants, including ECD listing and the kidney ‘Minimum Acceptance Criteria’ (MAC). The kidney MAC are provided annually, apply universally to all of a program's candidates, and are used for postmatch run screening only for offers placed through the UNOS organ center.

Our analysis has several limitations, notably that due to the limitations of our data set we were unable to compare some criteria to center practice. Our analysis of surplus offers treated a center's behavior as identical for all patients; although we showed that it is common for centers to use the same value for most patients, our approach did not account for such patient-level variation as exists. Although we believe that underutilization of the criteria system leads to increased CIT and discard, the effect is not concentrated in underutilizing centers but rather across the whole system. Furthermore, it is not possible to accurately quantify this effect. In using Poisson regression for modeling the relative rate of acceptance for adult versus pediatric registrants, we were forced to adjust the variance either to account for the imperfect fit of binary data to a Poisson model, or for center-level clustering; we chose the latter course. Nevertheless, although our confidence intervals may be artificially narrow, the consistent trend across many different criteria is clear.

Mortality while waiting for a transplant organ is high, and due to the shortage of organs, our community has an imperative to distribute organs as efficiently as possible. A system to direct offers of marginal organs to centers willing to accept them is a worthy and important endeavor. However, due to underutilization, overly broad criteria, and the univariate nature of the criteria, the system as currently configured does not appear to be meeting that goal.

Footnotes

  • 1

    UNet is a service mark of UNOS.

  • 2

    DonorNet 2007 is a registered trademark of UNOS.

Acknowledgments

We report an analysis of data collected by the Organ Procurement and Transplantation Network (OPTN). The OPTN is supported by Health Resources and Services Administration contract 234-2005-370011C. The analyses described here are the responsibility of the authors alone and do not necessarily reflect the views or policies of the Department of Health and Human Services, the OPTN, or UNOS; nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation. This study was not funded in any way by a commercial organization.

Ancillary