The contents of this article are the responsibility of the authors alone and do not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government.
This work was supported in part by the National Institutes of Health (grant T32 DK07634 to Eric S. Orman; grant 1KL2-RR025746 to A. Sidney Barritt IV; grant 1-K12 HS019468-01 to Stephanie B. Wheeler; and grant UL1-RR025747 to Eric S. Orman, A. Sidney Barritt IV, and Stephanie B. Wheeler) and by the Health Resources and Services Administration (contract 231-00-0115).
Liver transplantation is the treatment of choice for patients with end-stage liver disease, but utilization is limited by inadequate organ availability.1 Efforts to expand the donor pool have included living donation, split liver transplantation, and the use of extended criteria donors, which include older donors, donors with fatty livers, and donation after cardiac death (DCD) donors.1
Although the intention of extended criteria donation is to increase transplantation, posttransplant outcomes can be inferior when such organs are used.2-8 DCD grafts, procured after the donor's circulation ceases, are subject to a period of poor perfusion during the asystolic phase, which can lead to ischemic bile duct damage.9 Consequently, they carry an increased risk of recipient morbidity and mortality in comparison with standard donation after brain death (DBD) grafts, which are procured while circulation is supported.4, 7 DCD transplants are also associated with longer hospital stays and increased costs,10 and they may not be cost-effective in comparison with a strategy of remaining on the transplant waiting list until a DBD liver is available.11 Advanced donor age and graft steatosis (fatty liver) are also associated with graft failure.2, 3, 5 For these reasons, transplant centers are more likely to reject liver offers from DCD donors, older donors, and donors with liver steatosis.12, 13
After years of liver transplantation growth in the United States, the annual number of deceased donor transplants has fallen since 2006.14 The reasons for this decline are unclear but are not entirely due to stagnant donation rates because the decline in liver transplantation has exceeded the decline in organ donors.14 One potential reason for the decline is increasing donor liver nonuse due to unacceptable organ quality.15 As the population ages16 and becomes increasingly obese,17 increasing extended criteria donation could contribute to increasing nonuse. It also remains unclear whether the attempt to increase the number of available livers through increased DCD is being undermined by the increased nonuse of such livers.
Organ donors from whom at least 1 organ is used for transplantation are of particular interest because they lack an absolute contraindication to organ donation, such as an infectious condition that could be transmitted to the recipient. Liver nonuse within this group of successful organ donors is, therefore, likely due to particular issues with liver quality. We sought to evaluate the association between various donor characteristics and liver nonuse in this subgroup of organ donors in the United States and to examine temporal trends in characteristics that may explain the decline in liver transplants.
PATIENTS AND METHODS
Data Source and Study Population
We used Standard Transplant Analysis and Research files from the United Network for Organ Sharing (UNOS); these files contain information on all deceased donors in the United States since October 1, 1987. We limited the sample to donors ≥ 18 years old from whom at least 1 organ (liver, heart, intestine, kidney, lung, or pancreas) was transplanted into a patient. Split liver donations were excluded because of the likely differences in the liver quality required for split livers versus whole livers. Donors with a body mass index (BMI) < 14 or > 50 kg/m2 were also excluded because these data were likely the result of data entry errors.
The primary outcome was donor liver nonuse, which was defined as a liver that was not procured or a liver that was procured but not transplanted.
The key variables of interest were age, BMI, diabetes, hypertension, and DCD. BMI, diabetes, and hypertension were chosen because of their associations with nonalcoholic fatty liver disease (NAFLD).18 Liver histology was available for only 30% of the donors and was, therefore, not included in the main analyses. It was included in a secondary analysis as a dichotomized variable (macrosteatosis: <30% versus ≥30%).19 Because the relationship between age and organ nonuse was not linear, we specified ages categorically (<30, 30-39, 40-49, 50-59, and ≥60 years); similarly, we specified BMIs categorically (<18.5, 18.5-24+, 25-29+, 30-39+, and ≥40 kg/m2).
We examined additional donor factors that could be associated with the decision to use a liver.5, 13 We considered the following: year of organ recovery (continuous), sex, race as determined by the organ procurement organization (white, black, or other), cause of death [trauma, cerebrovascular accident (CVA), or other], lifetime smoking (>20 pack-years), alcohol use (≥2 drinks per day), history of illicit drug use, presence of infection, inotrope use at cross-clamping, history of malignancy, alanine aminotransferase (ALT), bilirubin, hepatitis C antibody, hepatitis B core antibody, blood type, and UNOS region. Within the subgroup of DCD donors, a prolonged warm ischemia time (WIT; >20 minutes) was also evaluated.20 The ALT and bilirubin levels were not normally distributed, and they were categorized as ≤40, 41 to 200, 201 to 400, and >400 U/L and as ≤1.2, 1.3 to 2.5, 2.6 to 5, and >5 mg/dL, respectively. Aspartate aminotransferase was not included because of its collinearity with ALT. Donor height, a component of the donor risk index,5 was not included because it is also a component of BMI.
Time Periods and Missing Data
Trends in donor characteristics and liver utilization were examined from 1988 to 2010. All other descriptive and multivariate analyses were restricted to recoveries performed between June 30, 2004 and December 31, 2010. We selected this time period for our analyses because of the completeness of the UNOS data and our interest in recent causes of liver nonuse. In particular, alcohol intake was not systematically recorded before this time. BMI, diabetes, hypertension, DCD, and laboratory values were available since 1995 with varying degrees of completeness. However, from 2004 onward, each individual variable of interest was missing for <3% of the donors. All variables were present for 92% of the donors.
Categorical variables were reported as proportions and counts. Continuous variables were reported as means and standard deviations when they were normally distributed and as medians and ranges otherwise. Bivariate comparisons between categorical variables were performed with Pearson's χ2 test. Comparisons between continuous variables were performed with the Student t test or the Wilcoxon rank-sum test as appropriate. Multivariate logistic regression was used to examine associations between donor characteristics and liver nonuse, controlling for all independent variables. Because transplant center practices can change over time, we assessed potential interactive effects between the year and all other variables with the likelihood ratio test. For instance, we hypothesized that centers may be more likely to use livers from older donors over time because the population is aging and emphasis has been placed on extended criteria donors nationally.21 Interaction terms for age and DCD by year were statistically significant, improved the overall model fit, and were included in the final model. Other interaction terms were omitted from the final model estimations.
Subgroup analyses were performed with additional logistic regression models. For the 30% of donors for whom histology results were available, the association between macrosteatosis and nonuse was evaluated. To explore trends in the impact of DCD on utilization in individual UNOS regions, a model was constructed that included interaction terms for UNOS regions with DCD. To determine whether there were differences in odds ratios (ORs) for donor characteristics between the DBD and DCD subgroups, interaction terms between DCD and all other factors were included in another model. Within the subgroup of DCD donors, the association between prolonged WIT and nonuse was also determined with logistic regression, with adjustments made for all other independent variables.
To determine the relative impact of each factor on the overall nonuse of livers, population attributable risk proportions (PARPs) were calculated.22 Despite being labeled proportions, PARPs for various donor characteristics do not sum to 100% because the characteristics overlap and the PARPs are not independent. Separate PARPs were calculated by year to illustrate temporal trends. Variance-weighted least squares linear regression was used to test trends in PARPs.
All P values were based on 2-sided tests, and P values < 0.05 were considered statistically significant. Analyses were performed with Stata 12 (StataCorp LP, College Station, TX). The institutional review board of the University of North Carolina approved this study.
ALT, alanine aminotransferase; BMI, body mass index; CI, confidence interval; CVA, cerebrovascular accident; DBD, donation after brain death; DCD, donation after cardiac death; NAFLD, nonalcoholic fatty liver disease; OR, odds ratio; PARP, population attributable risk proportion; UNOS, United Network for Organ Sharing; WIT, warm ischemia time.
Study Population Characteristics
Between January 1, 1988 and December 31, 2010, 107,259 individuals meeting inclusion and exclusion criteria donated at least 1 organ; 41,503 (38.7%) donated after June 30, 2004. After this date, livers were used for transplantation in 33,895 patients (81.7%), and 7608 livers (18.3%) were unused. The liver was used from 89.6% of successful heart donors, from 99.7% of intestine donors, from 78.4% of kidney donors, from 90.2% of lung donors, and from 96.9% of pancreas donors. When more than 1 organ was used, 95% of the livers were used. When only one organ was used, 51.2% of these were livers; 46.5% were one or both kidneys; and hearts, intestines, lungs, and pancreas each comprised < 2%. The mean donor age was 43.7 years, 40.9% were female, and 68.4% were white (Table 1). The mean BMI was 27.4 kg/m2, 10.9% had diabetes, and 36.3% were hypertensive. Nine percent were DCD donors, 18.4% had a history of alcohol use, 34.0% had a history of drug use, 3.8% were hepatitis C antibody–positive, and 42.2% had an ALT level > 40 U/L. Among those with available histology, 13.7% of those with histology available had ≥30% macrosteatosis.
Table 1. Characteristics of Organ Donors: Overall and According to the DCD Status (2004-2010)
Temporal Trends in Donor Characteristics and Liver Utilization
From 1988 to 2010, the mean donor age increased from 34.6 to 43.3 years. During that time, the percentage of donors who were 18 to 39 years old decreased from 65.6% to 40.3%, whereas the percentage of donors who were 50 years old or older increased from 16.0% to 38.0% (Fig. 1A). In 1988, 84.9% of the donors were white, whereas 67.1% were white in 2010. The prevalence of obesity (BMI ≥ 30 kg/m2) increased from 15.0% in 1995 to 30.3% in 2010 (Fig. 1B). The prevalence of diabetes and hypertension increased from 3.2% and 22.6% in 1995 to 12.0% and 37.2% in 2010, respectively (Fig. 1C,D). In 1995, 65.1% of the donors had an ALT level < 40 U/L, whereas 56.2% had an ALT level < 40 U/L in 2010. The bilirubin level, however, did not change substantially. The number of DCD donors increased from 44 (1.1%) in 1995 to 730 (11.2%) in 2010 (Fig. 2). Since 2006, the rise in DCD has been accompanied by a proportionate decline in DBD donation. The proportion of livers not used fell between 1988 and 2004 from 66.2% to 14.8% and then increased to 20.7% (Fig. 2). This utilization pattern was seen across UNOS regions, even in region 9, which traditionally has been the most aggressive user of expanded criteria livers (Fig. 3).23 From 1995 to 2010, the proportion of donors from whom only 1 organ was used increased slightly from 25.8% to 30.6%. In each year, the median number of organs used was 2. Within the group of patients from whom only 1 organ was used, 25.6% of the organs were livers in 1995. This figure increased to 59.1% in 2004 and then fell to 45.5% in 2010. Conversely, 67.2% were kidneys in 1995. The proportion of kidneys fell to 38.2% in 2004 and then increased to 52.2% in 2010.
Characteristics of DCD Donors
DCD donors (mean age = 41.3 years) were on average younger than DBD donors (mean age = 43.9 years, P < 0.001; Table 1). The mean BMIs for DCD and DBD donors were 27.7 and 27.3 kg/m2, respectively (P < 0.001). Diabetes and hypertension were more common in DBD donors. DCD donors were more likely to be white, to have a history of drug or alcohol use, and to have an ALT level > 40 U/L. DBD donors were more likely to be female, to have CVA as a cause of death, to have elevated bilirubin levels, and to be exposed to viral hepatitis. DBD donors were also more likely to receive inotropic support. For DCD, the median WIT was 15 minutes (interquartile range = 10-24 minutes). The WIT was prolonged (>20 minutes) in 34.3%. The prevalence of prolonged WIT fell from 41.1% in 2002 to 26.7% in 2005 and then returned to 41.1% in 2010.
Liver Nonuse According to Donor Characteristics
In bivariate analyses, nonuse was associated with increasing donor age up to 60 years, above which the liver was more likely to be used. In comparison with normal-weight donors, donors with higher and lower BMIs were less likely to be used. Diabetes and hypertension were both associated with nonuse. The proportions of nonuse were 58.2% for DCD livers and 14.4% for DBD livers. Livers from black donors were more likely to be used. Nonuse was also associated with smoking, alcohol use, increasing ALT and bilirubin levels, and exposure to viral hepatitis. As for livers from donors with histology available, 9.4% were not used when macrosteatosis was <30%, and 52.9% were not used when macrosteatosis was ≥30%. Within the DCD subgroup, 67.9% of the livers were not used when the WIT was >20 minutes, and 52.9% were not used when it was ≤20 minutes.
In the multivariate analysis, liver nonuse was associated with increasing donor age up to 60 years (Table 2). As in the bivariate analysis, livers from donors 60 years old and older were more likely to be used than livers from donors who were 50 to 59 years old. The strength of the association between age and liver nonuse increased over time (P for interaction = 0.002). DCD was strongly associated with liver nonuse, and this relationship became more pronounced over time. The OR for DCD increased from 5.53 [95% confidence interval (CI) = 4.57-6.70] in 2004 to 21.31 (95% CI = 18.30-24.81) in 2010 (P for interaction < 0.001). High and low BMIs and diabetes were associated with nonuse, but hypertension was not (Table 3). Black race and illicit drug use were independently associated with organ use, whereas female sex, other race, CVA as a cause of death, alcohol use, increasing ALT and bilirubin levels, and hepatitis C antibody were associated with nonuse. Macrosteatosis ≥ 30% was strongly associated with nonuse among those with histology available (adjusted OR = 11.16, 95% CI = 9.75-12.77). The ORs for most factors were similar in DBD and DCD donors. The strength of the associations between nonuse and an age of 30 to 49 years, alcohol use, nontraumatic causes of death, and elevated ALT levels were greater for DBD donors versus DCD donors (Table 4). In contrast, an age ≥ 50 years was more strongly associated with nonuse in DCD donors. Among DCD donors, prolonged WIT was associated with nonuse (adjusted OR = 2.06, 95% CI = 1.71-2.47).
Table 2. Trends in Adjusted ORs for Liver Nonuse by Donor Age and DCD
The ORs are based on a multivariate logistic regression model adjusted for all other variables in this table as well as the age, DCD status, UNOS region, and ABO blood type.
Cause of death
Illicit drug use
Hepatitis C antibody
Hepatitis B core antibody
Table 4. Adjusted ORs for Liver Nonuse for Donor Characteristics That Differ by the DCD Status
DBD [OR (95% CI)]
DCD [OR (95% CI)]
NOTE: The ORs are based on a multivariate logistic regression model adjusted for all variables in Tables 2 and 3 as well as the UNOS region and ABO blood type. The interaction terms for the listed variables by DCD were significant; the ORs for the DBD and DCD subgroups are presented.
Age in 2010 (years)
Cause of death
Population Attributable Risks for Liver Nonuse
DCD accounted for 8.7% of liver nonuse in 2004; this proportion increased to 28.4% in 2010 (Fig. 4). PARPs for DCD increased over time in all regions except region 6, in which the PARP peaked at 18.8% in 2006 and fell to 15.3% in 2010. The proportion of nonuse due to an age ≥ 50 years increased from 16.7% in 2004 to 22.9% in 2010. No other factor increased its impact on organ nonuse by more than 5% during this time. In 2010, a BMI ≥ 25 kg/m2 explained 17.5% of nonuse, but diabetes explained only 1.0% (Table 5). White race (versus black race) was responsible for 16.8% of nonuse. Alcohol use accounted for 10.9% of organ nonuse, an ALT level > 40 U/L accounted for 19.9%, a bilirubin level > 1.2 mg/dL accounted for 9.9%, and hepatitis C antibody accounted for 2.3%. The PARP for prolonged WIT within the DCD subgroup remained unchanged during the study period (8.1%, P for trend = 0.36).
Table 5. Overall Liver Graft Nonuse Due to Donor Characteristics in 2010
NOTE: The percentages were calculated as PARPs based on a multivariate logistic regression model.
The percentages do not sum to 100 because PARPs are not independent.
Cause of death
Hepatitis C antibody
The recent decline in the number of liver transplants performed in the United States is multifactorial. One possible reason for this decline is a decrease in donor organ quality leading to a decrease in utilization. In this study of all donors in the United States from whom at least 1 organ was transplanted, we found that the proportion of livers not used reversed its downward trend in 2004 and rose through 2010 (Fig. 2). In 2004, 841 of 5680 donated livers (14.8%) were not used; in 2010, 1345 of 6506 donated livers (20.7%) were not used. Had the rate of nonuse remained stable at 14.8%, 382 more livers from this group would have been transplanted in 2010.
Older donor age, higher BMI, and diabetes are independently associated with liver nonuse. As in the general population,16, 17 all these factors have become more common in the donor population over a long time period (Fig. 1). Therefore, these findings suggest that the quality of donated livers has been declining for many years. Despite this deteriorating quality, nonuse continued to decrease before 2004, possibly as a result of the gradual expansion of extended criteria donor liver use while acceptable posttransplant outcomes were maintained.24 However, the tolerance for worse outcomes with extended criteria donor livers has probably reached a limit, and this has led to declining utilization since 2004.
DCD is also independently associated with liver nonuse. In contrast to age, BMI, and diabetes, the increase in the prevalence of DCD and the nearly 4-fold increase in the odds of DCD livers not being used are particularly remarkable. This increasing reluctance to use DCD livers is undoubtedly the result of the growing recognition that posttransplant outcomes are worse, and it may be compounded by increased Medicare scrutiny of transplant programs' outcomes.25 Such reluctance is further justified by recent data demonstrating that patients with low Model for End-Stage Liver Disease scores and those with hepatocellular carcinoma exception points may be better served by declining DCD organs and waiting for a DBD liver.11 Although a prolonged WIT was associated with nonuse among DCD livers, this association did not translate into a rising PARP for prolonged WIT alone, perhaps because of fluctuating rates of prolonged WIT over time. Despite the relatively recent growth of DCD, it appears to be overtaking the other factors as a common reason for liver nonuse. By 2010, DCD accounted for more than one-quarter of unused livers. The increasing impact of DCD and decreasing utilization were seen across UNOS regions and were not the result of an outlier region influencing national trends. Even region 9, which historically has been significantly more likely to use marginal-quality livers,23 showed a similar increase in nonutilization since 2004 in comparison with other regions (Fig. 3).
Others have raised concerns about the increasing reliance on DCD. In the Netherlands, DCD increased, whereas DBD decreased; this resulted in no net gain of grafts for kidney transplantation.26 A 2005 conference took note of this trend and compared it to the United States experience, in which increases in DCD were accompanied by increases in DBD. They concluded that “the evolving DCD practice is expected to result in an absolute increase of organ donors, i.e., in addition to DBD.”27 Refuting this prediction, our results show that DBD has declined, whereas DCD has increased (Fig. 2).
A worrisome possible explanation for this pattern is that the withdrawal of cardiovascular support preceding DCD is occurring for donors who would have eventually progressed to brain death, so potential DBD donors are being converted to less desirable DCD donors. The reasons for possible conversions are unclear. There may be a growing preference for DCD among donor families and intensive care providers because they have become more educated on the DCD option. The Organ Donation Breakthrough Collaborative may have reinforced this preference by emphasizing DCD as a method for expanding the donor pool,28 particularly because DCD kidney transplant outcomes are comparable to DBD kidney transplant outcomes.29 Indeed, we found that among donors from whom only 1 organ was used, liver use has been declining since 2004, whereas kidney use has been increasing. Alternatively, the aggressive neurological management of donors may prevent the development of brain death.30 Such intensive neurological management has become more common and is associated with DCD.31
DCD has become more popular in recent years as a means of capturing organs that would not otherwise be available. The challenge is discerning between these situations of net organ gain and DBD-to-DCD conversions, in which the net gain of donated livers is nil. Unfortunately, our findings suggest that conversions could explain the rise in DCD because DBD has declined. Such conversions would be neutral in terms of the number of transplanted livers were it not for the increasing reluctance to use DCD livers. The transplant community must consider ways to capture data on any potential conversion problem. If such a problem exists, its harm to liver utilization could be vast.
Although the OR for not using a liver because of older age was less than the OR associated with DCD, the overall impact of age remains greater because advancing age is more common than DCD, and even a modest increase in age (≥30 years) is associated with nonuse. The lower OR for donors who are 60 years old or older versus donors who are 50 to 59 years old has been reported previously,13 and it likely reflects a selection bias for only the healthiest older individuals being considered for donation. Like the impact of DCD, the impact of age on liver nonuse is increasing, although less rapidly. Obesity is also becoming more prevalent among donors, but the overall impact of obesity has not increased, likely because of the relatively modest ORs for overweight and obese donors. Nevertheless, the population-level impact of obesity on liver nonuse may increase in the future as its prevalence increases.
Elevated donor ALT levels are becoming more common, are associated with liver nonuse, and have a substantial impact on overall utilization. The greatest increase in ALT prevalence was among donors with a mild-to-moderate elevation (41-200 U/L). Whether this finding represents another manifestation of NAFLD is unknown, but if true, it would also bode poorly for future organ quality. In contrast, black race was strongly associated with organ use in comparison with white race. The reasons for this difference are unclear, but they could be potentially related to the lower prevalence of NAFLD in blacks versus other races.32 This finding takes on particular importance because of the lower rates of organ donation in the black population.15
DBD and DCD donors represent 2 distinct populations, and one might expect that the relationships between donor factors and nonuse would differ between these groups. However, ORs for nonuse were actually similar for most donor factors. When they were different, most associations were attenuated in the DCD group, and this likely represented a selection bias for DCD donors deemed acceptable for the use of at least 1 organ.
Additional external factors could affect liver utilization. DonorNet is the electronic UNOS system used to facilitate organ placement since April 30, 2007. Through DonorNet, electronic organ offers can be made simultaneously to multiple recipient centers, and detailed donor information is available electronically to increase efficiency and improve utilization. Before the use of DonorNet, organ offers were made manually, and this limited communication and decision making in the donor-recipient matching process. After DonorNet was implemented, there was an increase in the percentage of recovered livers that were discarded, particularly among those with the highest donor risk indices.33 However, our findings show that utilization was already decreasing before DonorNet was implemented, so its effect on utilization remains unclear.
The main limitation of this study is the lack of details on reasons for the rise in liver nonuse. The reasons for organ nonuse are recorded broadly in the database as “poor organ function” or “biopsy findings.” The data set also lacked sufficient biopsy data to directly link donor steatosis to organ nonuse because biopsy data were available for only 30% of the donors. However, obesity and diabetes are strongly associated with NAFLD18 and have previously been used as surrogates for steatosis.13 Although we did find an association between steatosis and nonuse within the subgroup for whom histology was available, this represents a biased sample because donors who undergo liver biopsy are highly selected. Therefore, the estimate for this relationship cannot be generalized to the entire donor population, and the overall PARP was not calculated. Finally, this study does not address the stagnation in the rates of overall donation that is affecting the national volume of transplants.15 However, the utilization of available organs becomes even more critical if we cannot rely on expanding donation.
Despite these shortcomings, the UNOS database is well suited for addressing our study goal of reporting national trends in organ utilization. The database has complete coverage for all donors and organ dispositions nationwide. The misclassification of our primary outcome and DCD is probably rare and unlikely to affect our results. The misclassification of other variables such as diabetes may be more likely, but it is probably nondifferential and would, therefore, bias the results toward the null.
Among organ donors in the United States from whom at least 1 organ was used for transplantation, we found that the nonuse of livers fell until 2004 and has since increased. Concurrently, the prevalences of advanced donor age, elevated BMI, diabetes, and DCD have all increased. These factors are associated with liver nonuse, and the impact of DCD on nonuse has grown rapidly. A better understanding of the reasons for the increasing proportion of DCD in particular is critical to understanding this declining utilization. These trends, along with stagnant donation rates, suggest significant declines in liver transplant availability in the coming years.