The Comparative Effectiveness of Donation after Cardiac Death versus Donation after Brain Death Liver Transplantation: Recognizing Who Can Benefit

Abstract

Introduction

Due to organ scarcity and wait-list mortality, transplantation of donation after cardiac death (DCD) livers has increased. However, the group of patients benefiting from DCD liver transplantation is unknown. We studied the comparative effectiveness of DCD versus donation after brain death (DBD) liver transplantation.

Methods

A Markov model was constructed to compare undergoing DCD transplantation with remaining on the wait-list until death or DBD liver transplantation. Differences in life years, quality-adjusted life years (QALYs), and costs according to candidate Model for End-Stage Liver Disease (MELD) score were considered. A separate model for hepatocellular carcinoma (HCC) patients with and without MELD exception points was constructed.

Results

For patients with a MELD score <15, DCD transplantation resulted in greater costs and reduced effectiveness. Patients with a MELD score of 15 to 20 experienced an improvement in effectiveness (0.07 QALYs) with DCD liver transplantation, but the incremental cost-effectiveness ratio (ICER) was >$2,000,000/QALY. Patients with MELD scores of 21 to 30 (0.25 QALYs) and >30 (0.83 QALYs) also benefited from DCD transplantation with ICERs of $478,222/QALY and $120,144/QALY, respectively. Sensitivity analyses demonstrated stable results for MELD scores <15 and >20, but the preferred strategy for the MELD 15 to 20 category was uncertain. DCD transplantation was associated with increased costs and reduced survival for HCC patients with exception points but led to improved survival (0.26 QALYs) at a cost of $392,067/QALY for patients without exception points.

Conclusions

In conclusion, DCD liver transplantation results in inferior survival for patients with a MELD score <15 and HCC patients receiving MELD exception points, but provides a survival benefit to patients with a MELD score >20 and to HCC patients without MELD exception points.

INTRODUCTION

Organ scarcity remains a major challenge in the field of liver transplantation. Each year in the United States (US), over two thousand patients with end-stage liver disease (ESLD) die on the waitlist.[1] Livers from donation after brain death (DBD) donors are the largest source of grafts currently. Yet, brain death represents only a small fraction of all-cause mortality[2, 3], while cardiovascular death is the largest cause of mortality and could vastly expand the overall pool of donor organs.[2, 4]

Federal mandates from the Health Resources and Services Administration (HRSA) and Centers for Medicare and Medicaid Services (CMS)[5] have been successful in increasing the utilization of donation after cardiac death(DCD) livers.[1] However, outcomes after DCD liver transplantation are marred by higher complication rates, inferior survival, and higher costs compared to DBD liver transplantation.[6–12]

Prior studies have failed to identify which group of patients with ESLD might benefit from transplant with DCD grafts compared to remaining on the waitlist. In an analysis based on donor quality measured by the Donor Risk Index (DRI) score[13]and patient disease severity according to the Model for End-stage Liver Disease (MELD) score, a survival benefit was noted when patients with a MELD > 20 were given a high risk graft (DRI>1.65). [14] However, this analysis failed to distinguish between different types of “high risk” grafts, account for quality of life differences related to the high complication and re-transplantation rates, or consider the impact on costs.

This study considers both the cost and quality of life implications post-transplantation as well as outcomes beyond one-year survival to provide a more detailed understanding of the comparative effectiveness of DCD liver transplantation.

METHODS

Decision Analytic Model

A Markov model was utilized to compare two treatment strategies available to patients with ESLD: 1) transplantation with a DCD liver versus 2) remaining on the waitlist with the possibility of receiving a DBD liver according to the standard MELD-based allocation scheme. The model is based on the patient’s/transplant center’s perspective and is intended to identify the optimal choice for an individual patient according to their MELD score. The MELD score is a validated predictor of 90-day waitlist mortality (c-statistic=0.83).[15, 16] Consequently, the MELD score has been used to establish priority for liver transplantation since 2002. This model does not consider the impact of organ acceptance decisions on the overall waitlist cohort, but rather focuses only on maximizing the outcomes of individual patients. In our model, we included health states representing patients on the waitlist according to MELD score and post-transplant patients with (1) a functioning graft, (2) development of acute or chronic biliary complications, (3) all-cause graft failure resulting in re-transplantation, and (4) death. Costs, life-years, and quality-adjusted life years (QALYs) were calculated. Cycle length for the model was one month. A 10 year time horizon was chosen due to limitations of the available survival data probabilities. Half-cycle corrections and discounting for both costs and utilities at 3% were included. Figure 1 is a graphic illustration of the health states and transitions represented in the model. Incremental cost-effectiveness ratios (ICER) based on the differences in costs between the two treatment strategies divided by the differences in effectiveness as measured in QALYs are reported.

Figure 1.

Patient Population

A 50 year-old Caucasian male with ESLD was utilized as the index patient. Patients with fulminant liver failure were excluded, and patients were assumed to not have undergone transplantation previously. Patients were grouped and analyzed by MELD quintile (<11, 11–14, 15–20, 21–30, and >30). Those patients with hepatocellular carcinoma (HCC) including subgroups of patients with and without MELD exception points were considered in a separate analysis. Additionally, this analysis focuses on those individuals without a living donor option. This represents the majority of patients as more than 96% of liver transplants are performed using deceased donors.[1]

Model Inputs

Waitlist Probabilities

Probabilities of receiving a DBD transplant, dying on the waitlist, and disease progression according to MELD score were based on national data available from Table 9.2b of the 2007 Scientific Registry of Transplant Recipients (SRTR) annual report.[1] These probabilities are presented in Table 1. The 30-day probability of receiving a DBD transplant according to the MELD score was calculated for each region from monthly event counts for 2007–2009 provided by UNOS.

Table 1.

Waitlist probabilities used in model.^*

Starting MELD	Transplant (DBD)	Death	MELD progression
			<11	11–14	15–20	21–30	>30
<11	0.007	0.003	0.959	0.035	0.005	0.001	0
11–14	0.007	0.004	0.052	0.887	0.058	0.003	0
15–20	0.048	0.015	0.007	0.082	0.854	0.054	0.003
21–30	0.193	0.044	0	0.003	0.213	0.741	0.043
>30	0.476	0.206	0	0	0	0.302	0.698

^*probabilities from 2007 SRTR annual report.[1] All probabilities are 30 day probabilities.

Post-Transplantation Probabilities

The risks of post-transplant complications including early graft failure secondary to primary non-function and vascular complications and biliary complications were incorporated. Acute, resolvable biliary complications and chronic, unremitting ischemic cholangiopathy (IC) were included in the model. Acute biliary complications include biliary leaks and anastomotic strictures which often resolve after one or more therapeutic endoscopic, percutaneous, or operative procedures. IC is characterized by diffuse intra-hepatic stricture formation of a more chronic nature and is associated with a higher rate of re-transplantation. IC has been demonstrated to result in significant healthcare costs and impact on quality of life.[7, 10–12, 17] The probabilities for acute biliary complications and IC were derived from data published in a meta-analysis of eleven single-institution studies.[18] These complications are not adequately reported to the SRTR national registry; and as such, this represents the most comprehensive analysis currently available. Odds ratios (OR) from this meta-analysis were used to reflect the increased risk of complications for DCD transplants.[18]

Probabilities of death after transplantation and graft failure requiring re-transplantation were based on national SRTR standard analysis files. Post-transplant patient survival and re-transplantation rates were estimated utilizing life table survival analysis based on the SRTR data set for DCD (n=1,113) and DBD (n=42,254) liver transplants performed between 4/1/1996 and 8/1/2008[19]. Pediatric (age<18), multi-organ transplants (except simultaneous liver-kidney transplants), patients not undergoing primary transplantation, and patients with no follow up (n=297 [0.6%]) were excluded. SRTR data were also used to analyze survival after re-transplantation (n=5,153). Patients were censored at date of last known follow-up. Only one re-transplantation per patient was represented in the model. Actuarial survival based on age, sex, and race-specific estimates [2] was subtracted from post-transplant survival probabilities and modeled separately from disease-specific survival.[20] Baseline estimates for post-transplantation probabilities are listed in Table 2. Survival analyses were performed using StataSE10 (Stata Corp, College Station, TX).

Table 2.

Post-transplantation event probabilities.

Parameter	Estimate	Source	Comparison	Reference
DCD patient survival*
1 yr survival	82.4% (95% CI = 80.1%-84.6%)	SRTR national registry (4/1/96–8/1/08) DCD n=1,113 DBD n=42,254	79.7%	Abt 2004[4]
3 yr survival	71.2% (95% CI = 68.1%-74.1%)		72.1%	Abt 2004[4]
DBD patient survival*
1 yr survival	85.9% (95% CI = 85.5%-86.2%)	SRTR national registry (4/1/96–8/1/08) DCD n=1,113 DBD n=42,254	85.0%	Abt 2004[4]
3 yr survival	77.4% (95% CI = 77.0%-77.8%)		77.4%	Abt 2004[4]
Re-transplant patient survival*
1 yr survival	67.6% (95% CI = 66.1%-68.9%)	SRTR national registry (4/1/96–8/1/08) n=5,153	66.9%	UNOS/OPTN 2008
3 yr survival	59.0% (95% CI = 57.5%-60.5%)		55.5%	UNOS/OPTN 2008
Acute biliary complications
DBD	11.7% ± 0.8%	Meta-analysis by Jay et al[18] DCD n=275 DBD n=1627
DCD	OR = 1.03 (95% CI = 0.52 –2.04)
Ischemic cholangiopathy
DBD	3.5% ± 0.3%	Meta-analysis by Jay et al[18] DCD n=288 DBD n=1725
DCD	OR = 12.52 (95% CI = 4.50 –34.84)
Re-transplantation *
DBD	6.8% ± 0.1%	SRTR national registry (4/1/96–8/1/08) DCD n=1,113 DBD n=42,254	5.4%	Selck 2008[9]
DCD	14.7% ± 1.1%		13.6%	Selck 2008[9]

^*Probabilities used in model based on monthly transitions from time to event survival analysis of SRTR dataset. One and three year survival and overall re-transplantation rates are provided in table to enable comparison to previously published literature.

Costs

All costs in this analysis were based on direct medical care costs. Inpatient and outpatient costs accrued over the analysis period were included, but costs related to lost wages or other societal costs were not included. Costs for hospitalizations, procedures, medications, and follow-up were drawn from the available literature and are presented in Table 3. Organ acquisition costs (OAC) were not included. All costs were adjusted to 2008 dollars using the consumer price index for medical care/services from the US Bureau of Labor Statistics.

Table 3.

Costs and utilities used in model. All costs are reported in 2008 adjusted US dollars.

	Base	Range	Reference
Costs
Base cost of transplant (includes 30 day post-transplant period)	$90,035	$66,515–113,556	Salvalaggio, Axelrod, Northup, Bennett, Huang, Lin [17, 38, 40–43]
Incremental cost of transplant *			Salvalaggio, Axelrod [17, 40]
• According to MELD
• - REF: MELD <15
• - MELD 15–20	$7,517	$1,037–13,997
• - MELD 21–30	$24,953	$17,922–31,984
• - MELD > 30	$72,119	$66,056–79,183
• DCD #	$11,860	$5,129–18,590
• Re-transplantation	$44,476	$14,418–74,535
Cost of waitlist (monthly; visits and labs)	$180	$91–360	Northup, Bennett, Lin [38, 41, 43]
Incremental pre-transplant costs according to MELD (annual)			Buchanan [44]
• REF: MELD<15
• MELD 15–20	$10,700	$0–46,300
• MELD 21–30	$62,000	$23,400–100,600
• MELD >30	$145,500	$104,000–187,100
Healthcare costs in HCC patients (annual)	$31,740	$15,870–63,480	Lang [45]
Cost of post-transplant outpatient follow-up (monthly; visits and labs)	$306	$153–612	Bennett, Lin[41, 43]
Cost of transplant meds (annual)	$18,848	$4,294 – 37,574	Bennett, Shenoy[41, 46]
90 day cost of biliary complications	$15,810	$4,737–34,606	Northup, Ammori [38, 47]
Annual cost of ischemic cholangiopathy (readmissions, ERCP/PTC, meds)	$78,689	$33,520–123,859	Jay, Ammori [11, 47]
Cost of cirrhotic death	$42,297	$21,148 – 84,656	Lin[43]
Cost of non-cirrhotic death	$5,429	$1,086–10,858	Huang[42]
Utilities
Waitlist/End-stage liver disease
MELD < 15	0.73	0.50–0.90	Northup, Bennett, Bryce, McLernon, Sagmeister, Schecter, Younossi[38, 41, 48–52]
MELD 15–30	0.67	0.40–0.90
MELD >30	0.56	0.30–0.81
Post-transplant	0.83	0.62–0.87	Northup, Bennett, McLernon, Sagmeister, Schecter[38, 41, 49–51]
Post-op disutility (applied to initial 30 day post-operative period)	50%		Northup[38]
Biliary complications/Ischemic cholangiopathy	0.71	0.39–0.81	Parikh, Younossi, Bondini, Kim, Longworth, Olsson[52–57]

^*incremental cost applied in addition to base transplant cost.

^#incremental cost of DCD transplant based on costs identified for livers with DRI>1.8.[17]

Utilities

Table 3 also lists the utility values and ranges for the various health states represented in the model. A one-month utility toll was applied when patients underwent transplantation. There is currently no utility examining patients with biliary complications available. The utility used in the model for this health state is based on a recent analysis of QOL data in a small group of patients with IC and studies evaluating patients with cholestatic liver disease who experience symptoms of jaundice, pruritus, and cholangitis. All utilities represented in the model are summary estimates based on the available literature.

Sensitivity Analysis

A probabilistic sensitivity analysis was conducted for each initial MELD state using Monte Carlo second-order simulation based on the overall distributions for all parameter estimates (probabilities, costs, and utilities). Differences in graft quality amongst DCD and DBD grafts were evaluated in the Monte Carlo simulations. One thousand Monte Carlo iterations were performed selecting new random values for each parameter based on the distributions of these parameters.

One-way sensitivity analysis (varying single parameters individually) was used to identify threshold values for key variables including probability of receiving a DBD transplant while on the waitlist. We examined regional variation in the probability of DBD transplant according to MELD score. United Network for Organ Sharing (UNOS) data on transplant event counts and candidate registrations from 1/1/2003 to 12/31/2009 was used to determine average 30-day probabilities. Finally, expected value of perfect information (EVPI) analysis was utilized to further examine uncertainty in the model. Tree Age Pro 2009 (version 6.3.1.0, Williamstown, MA) decision analysis software was utilized for model development and analysis. This study was approved as exempt by the Institutional Review Board of the Feinberg School of Medicine at Northwestern University.

RESULTS

Cost-effectiveness results for the index case analyses according to the MELD score are summarized in Table 4. For patients with a MELD < 15, DCD transplantation was associated with both reduced effectiveness and greater costs compared to remaining on the waitlist. Patients with a MELD score 15–20 had a small improvement in effectiveness with DCD liver transplantation (DCD 4.43 QALYs versus DBD 4.36 QALYs). However, this increase (+ 0.07 QALYs) was associated with an increased cost of $150K (DCD $357K versus DBD $207K) resulting in an incremental cost-effectiveness ratio (ICER) of more than $2million/QALY. Patients with a MELD score of 21–30 or >30 also derived benefit from DCD transplantation (+0.25 and +0.83 QALYs respectively) which corresponded to ICER values of $478K/QALY and $120K/QALY, respectively. For HCC patients, DCD transplantation incurred increased costs and decreased survival (see Table 4). However, when stratified into HCC with and without MELD exception points, HCC patients without exception points derived a survival benefit (+0.26 QALYs) from DCD transplantation with an ICER of $392K/QALY.

Table 4.

Incremental cost-effectiveness results for base case analysis.

Strategy	Cost	Incr Cost	Eff	Incr Eff	Incr NMB	Incr C/E (ICER)
MELD <11
Stay on waitlist	$166K		4.86
DCD transplant	$357K	$190K	4.43	−0.43	−$233K	(Dominated)
MELD 11–14
Stay on waitlist	$180K		4.66
DCD transplant	$357K	$177K	4.43	−0.24	−$201K	(Dominated)
MELD 15–20
Stay on waitlist	$207K		4.36
DCD transplant	$357K	$150K	4.43	0.06	−$144K	$2,587,046
MELD 21–30
Stay on waitlist	$239K		4.18
DCD transplant	$357K	$118K	4.43	0.25	−$93K	$478,222
MELD >30
Stay on waitlist	$257K		3.59
DCD transplant	$357K	$100K	4.42	0.83	−$17K	$120,144
All HCC patients
Stay on waitlist	$223K		3.84
DCD transplant	$334K	$111K	3.79	−0.05	−$116K	(Dominated)
HCC with exceptions
Stay on waitlist	$262K		3.93
DCD transplant	$334K	$72K	3.79	−0.14	−$86K	(Dominated)
HCC without exceptions
Stay on waitlist	$234K		3.53
DCD transplant	$334K	$100K	3.79	0.26	−$74K	$392,067

Probabilistic Sensitivity Analysis

Probabilistic sensitivity analysis was performed for each MELD quintile. For patients with a MELD <11 or a MELD 11–14, DCD transplantation is dominated (increased costs, reduced QALYs) in 85% and 76% of the 1,000 iterations, respectively. For patients with a MELD 15–20, DCD liver transplantation was associated with increased QALYs in only 56% of the iterations, but again the ICER value was < $100K/QALY in only 0.6% of iterations. Increased effectiveness was observed in 82% of the iterations with DCD transplantation for MELD 21–30 patients, and an ICER < $100K/QALY was present in 12% of iterations. Finally in MELD >30 patients, increased QALYs were identified in 98% of cases and an ICER < $100K/QALY in 55%. An ICER < $50K/QALY was identified in 23% of iterations.

One-way Sensitivity Analyses

The impact on effectiveness imparted by the monthly probability of receiving a DBD liver and the monthly probability of death on the waitlist according to MELD score was examined utilizing one-way sensitivity analyses. In the MELD >30 cohort, DCD transplantation was associated with improved QALYs unless the monthly probability of receiving a DBD transplant exceeded 76% (mean national probability = 48% [1]). For the MELD 21–30 cohort, staying on the waitlist became the more effective strategy when the monthly probability of DBD transplant exceeded 32% (mean national probability = 19% [1]) Regional probabilities varied from 7% (region 1) to 26% (region 9)(Figure 2). Finally, a monthly probability of DBD transplant greater than 8% in patients with a MELD 15–20 was necessary for DCD transplantation to be the preferred strategy (mean national probability = 5%[1]).

Figure 2.

Figure 3 displays regional 30 day probabilities of receiving a DBD liver according to MELD score based on UNOS data. There is substantial regional variation in the 30 day probability of receiving a DBD transplant for patients with MELD>20. Region 1 had the lowest probability (9.6% ± 3.5%) while region 3 exhibited the highest probability (43.9% ± 7.0%) of receiving a DBD liver.

Figure 3.

Additionally, we compared the treatment strategies excluding quality-adjustment by removing the contribution of utilities. Excluding the utilities from our analysis produced similar results to our quality-adjusted analysis. The increased effectiveness and ICER for the MELD >30 group was 1.56 QALYs and $64K/life-year, respectively. The MELD 15–20 and 21–30 groups were associated with ICERs of $385K and $177K/life-year, and DCD transplant was dominated in the remaining MELD quintiles. Finally, we looked at the differences in the ICERs over shorter and longer time horizons and found minimal variability. For instance in the MELD >30 group we observed an ICER of $106K/QALY at 5 years and $130K/QALY at 20 years.

Expected Value of Perfect Information (EVPI)

Finally, we determined the value of more precise information regarding probabilities, costs, and utility estimates employed in the model. The EVPI represents the expected cost of uncertainty or expected opportunity lost due to not having precise estimates of probabilities, costs and utilities. Unlike a conventional sensitivity analysis, it incorporates both the probability that a decision may be wrong and the consequences of a wrong decision. Consequences are measured in terms of incremental net benefit, equal to incremental life years at $100K/QALY minus incremental cost. Incremental net benefits are shown in Table 4. The EVPI for all probabilities, costs, and utilities in our model ranged from a high of $21,890 for MELD >30 group to values of than $2670 and $930 for the MELD 21–30 and MELD 15–20 groups, respectively. These values are insignificant compared to the tens to hundreds of thousands of dollars of incremental net benefits shown in Table 4 indicating that even acquiring perfect information of probabilities, costs and utilities would be unlikely to alter the choice between immediate DCD transplant and remaining on the waitlist. In short, there is a clear choice in terms of incremental net benefit.

DISCUSSION

Given the dramatic growth in healthcare costs and the current national debate on value-based healthcare, a comparative effectiveness focus has become increasingly prominent. The 2009 Institute of Medicine (IOM) report identified comparative effectiveness research (CER) as a top priority.[21] CER “provides an opportunity to improve the quality and outcomes of healthcare by providing more and better information to support decisions by the public, patients, caregivers, clinicians, purchasers, and policy makers.”[21] Moreover, cost-effectiveness analysis enables decision makers to allocate scarce resources in a manner that can maximize the overall health benefit for society.[22] Historically, an ICER less than $50K/QALY has been considered cost-effective according to a benchmark paper examining the costs of chronic hemodialysis.[23, 24] This study when adjusted according to the recent consumer price index would yield a willingness-to-pay threshold greater than $75–100K/QALY.

This analysis is the first attempt to evaluate the comparative effectiveness of DCD liver transplantation. Nowhere is the issue of resource scarcity more evident than in the field of organ transplantation. The increasing use of DCD livers was intended to mitigate organ scarcity, but evidence of its effectiveness has not been rigorously assessed. Concerns over the increasing use of DCD liver transplantation have risen with the publication of multiple studies demonstrating that these grafts have been associated with higher complication rates, inferior graft survival, and higher costs compared to DBD liver transplantation.[6–11, 17, 25] Most notable is the markedly increased incidence of ischemic cholangiopathy, a chronic diffuse biliary disorder, identified in 14–50% of DCD recipients (compared to 1–2% of DBD recipients).[6, 7, 10, 26, 27] Patients with IC are afflicted by jaundice, disabling pruritus, and cholangitis resulting in frequent readmissions and invasive biliary tract procedures with considerable quality of life implications.[10] Moreover, DCD recipients require re-transplantation more than twice as often leading to greatly increased costs. Finally, recent declines in DBD donation in the setting of overall gains in DCD donation in the past decade raised additional concerns about the potential “cannibalization” of DBD grafts.[1, 3, 28] As such, appropriate utilization of these grafts is of the utmost concern.

This study demonstrates that for patients with MELD scores < 15, DCD transplantation is associated with both higher costs and reduced effectiveness compared to remaining on the waitlist. For patients with a MELD >20, there is an increase in effectiveness after DCD transplantation in essentially all iterations performed in the Monte Carlo sensitivity analysis. The QALY benefit for MELD 15–20 patients remains much less clear, with 56% of iterations demonstrating a benefit from DCD transplantation. These findings confirm prior research by Merion et al. which demonstrated a survival benefit with liver transplantation compared to remaining on the waitlist for patients with a MELD >15.[29] Subsequently, Schaubel at al. established that only patients with a MELD >20 derived a benefit when transplanted with high DRI organs when donor quality was added to the equation.[14]

However, DCD transplantation according to recent practices was not a cost-effective strategy for any MELD group studied. Patients with a MELD >30 had the lowest ICER value at $120K/QALY. Additional consideration of healthcare costs can be a highly controversial topic because it goes beyond making healthcare decisions based solely on maximizing outcomes. Indeed, cost-effectiveness is not the focus of physicians when advising patients. However, given the limited liver organ resources available, these results are relevant for policy makers when establishing allocation and resource guidelines.

Additionally, the impact of local and regional variation on the probability of receiving a transplant while on the waitlist is an important topic. We identified a substantial amount of variation when the probabilities of receiving a transplant were examined across the eleven UNOS regions with some regions having a 30-day DBD transplant probability for patients with MELD>20 that was more than four-fold higher than other regions. While examining such variation is informative, it is limited in its ability to convey accurately the true variation occurring at the donor service area (DSA) and transplant center levels. According to UNOS data, the median time to transplant varies widely from one DSA to the next within a region. For example, in region 11 the median time to transplant ranged from 0.5 months in one DSA to >60 months in another.[30] However, there are insufficient data in the registry as many DSAs have recorded neither transplant events nor waitlisted patients particularly among the higher MELD strata making it impossible to derive accurate 30 day transplant probabilities according to DSA even when many years of data are aggregated.

To address the impact of variation in DBD availability in our model, one-way sensitivity analyses were performed evaluating the probability of receiving a DBD transplant. For MELD >30 patients, a 30-day probability of DBD transplant greater than 76% (compared with a national estimate of 48%) was necessary for worse survival with DCD transplantation. Similar differences in the probabilities in receiving a DBD transplant were identified for the other MELD quintiles. The threshold probability was again larger than the national estimate for MELD 21–30 (32% versus 19%) and MELD 15–20 (8% versus 5%). These data highlight the impact of the contextual environment on decision-making by clinicians.

Similarly, we examined the impact of recipient age at transplant by varying survival probabilities for ages 30–70 (data not shown). This yielded no differences in which strategy was more effective for a given MELD range or relative cost-effectiveness. Additionally, we also examined separate cohorts of HCC and HCV positive candidates. DCD was the inferior strategy for HCC patients overall and for those with exception points specifically resulting in increased costs and decreased survival. However, HCC patients not receiving exception points did have a survival benefit with DCD transplant at an ICER of $392K/QALY. HCV status had no impact on the model output yielding similar results to the MELD-based analysis for ESLD patients as a whole. These findings are consistent with other recent studies examining HCV.[31–34]

Limitations

There are several limitations to this analysis. First, the accuracy of the model output is directly dependent on the ability to estimate the “true” probabilities, costs, and utilities. A strength of this manuscript is that all model probabilities are derived from national registry data and a meta-analysis and do not rely on single institution results. To address questions related to the potential variability or inaccuracy of model parameters, probabilistic sensitivity analyses were performed varying all model parameters simultaneously according to their distributions. These analyses demonstrated significant stability of the findings for the upper and lower MELD quintiles. However for patients with a MELD score of 15–20, there was more uncertainty in whether DCD transplantation resulted in an improvement in QALY’s according to the probabilistic sensitivity analysis. Moreover, the extremely low EVPI for all probabilities, costs, and utilities indicates that resolving the uncertainty in these variables would be unlikely to change the conclusions of the study.

Similarly, the validity of utilities has been challenged based on concerns regarding participants’ abilities to accurately understand and assess the hypothetical situations proposed in typical standard gamble and time-trade-off scenarios. No utilities were available in the literature for patients with biliary complications following transplantation. The utility used in this model was based on QOL assessment of a small cohort of DCD recipients with IC. According to the QOL data collected on this small group of patients, the utility calculated was 0.55. Additionally, we included the range of utilities derived from studies in patients with native biliary tract disease. Ultimately, we chose a more conservative estimate of 0.71 for the utility associated with biliary complications providing a relative advantage and potential bias in favor of DCD transplant. However, the lower utility estimate of 0.55 was included in the larger range used in our probabilistic sensitivity analyses. Additionally, we examined the impact of the biliary complications utility in a one-way sensitivity analysis. For the MELD>30 group, DCD transplant was associated with increased effectiveness with ICERs ranging from $107K-191K/QALY. For the MELD 15–20 and 21–30 group, DCD transplant became dominated (a more costly and less effective strategy) at a biliary complication utility less than 0.65 and 0.46, respectively. Finally, we also examined pure effectiveness of the treatment strategies in the absence of quality adjustment which rendered similar results to our base-case analysis.

Next, we examined regional variations in waitlist probabilities. However, an additional limitation involves the impact on the waitlist in relation to increases or decreases in the number of available DCD livers. Augmenting the number of DCD liver transplantations nationwide would potentially have an impact because patients would be removed from the list at a faster rate and subsequently the probability of receiving a DBD transplant at a lower MELD score would improve. However, more recently, due to concerns over worse outcomes of DCD livers, there has been nearly a 10% decline in DCD liver utilization in the US[35] despite an overall increase in the supply of DCD livers over the last decade.[1] An additional benefit from DCD transplantation may have been observed if the impact of increasing the supply of available grafts on the waitlist as a whole had been considered in our model. Allocation policy reforms to address those patients disadvantaged with a failing DCD graft due to IC may be necessary to remedy this decay in DCD liver utilization.[36, 37]

Another limitation in this analysis is the exclusion of OAC which can differ greatly between DCD and DBD graft types. Costs of an individual donation can be greatly reduced by avoiding the time-consuming and expensive tests and treatment to allow a patient to progress to brain death seeming to favor DCD donation. However, this fails to take into account the higher non-recovery and discard rates that have been identified for DCD donors.[1] These factors result in more frequent trips by organ procurement teams in which organs were not obtained for transplant and higher standard acquisition costs for livers. Further research is needed to elucidate organ acquisition cost differences. Similarly, indirect costs including lost wages or other societal costs were not included. Given higher complication rates associated with DCD liver transplantation, inclusion of these costs would likely result in even more unfavorable ICER values for DCD transplants.

Additionally by choosing a ten-year time horizon, gains associated with prolonged survival after transplantation are potentially minimized. However, this base time horizon was selected due to concerns about the ability to extrapolate survival data beyond this time period, especially in light of the significant evolution in transplant practices. We did further confirm consistency of our findings by evaluating differences in the ICER according to 5, 10 and 20 year time horizons.

Finally, living donor liver transplantation was not considered as a therapeutic alternative in this model. Recent research has suggested that living donors represent a cost-effective option compared with medical management ($36K/QALY) and provide a survival benefit, but at a cost of $107K/QALY compared with standard deceased donor liver transplantation.[38] DCD liver transplantation has been increasingly considered for those recipients lacking an available living donor.[38, 39] By focusing our study on those individuals without a viable living donor option, we aimed to address the relevant question facing the majority of patients as living donor transplantation is still only performed for a small number of liver failure patients (currently ~3.5% of annual liver transplants.)[1]

Conclusions

For patients with a MELD <15, DCD liver transplantation was associated with reduced effectiveness and greater costs compared to remaining on the waitlist. Patients with a MELD score of >20 had higher QALYs with DCD liver transplantation, but improvements in QALYs for MELD 15–20 patients were not stable when model parameters were varied. For HCC patients with exception points, DCD transplantation was dominated resulting in increased costs and decreased survival. However, a survival advantage was seen with DCD transplantation for HCC patients not receiving exception points. Patients with a MELD > 30 had the lowest ICER value at $120K/QALY. Given the current focus on comparative effectiveness, the lack of cost-effectiveness of DCD liver transplantation according to recent practices needs to be considered in making decisions regarding the utilization and allocation of DCD livers.

ABBREVIATIONS

US

United States

ESLD

End-Stage Liver Disease

HRSA

Health Resources and Services Administration

CMS

Centers for Medicare and Medicaid Services

DCD

Donation after Cardiac Death

DBD

Donation after Brain Death

DRI

Donor Risk Index

MELD

Model for End-stage Liver Disease

QALY

Quality-Adjusted Life Years

ICER

Incremental Cost-Effectiveness Ratio

SRTR

Scientific Registry of Transplant Recipients

IC

Ischemic Cholangiopathy

OR

Odds Ratio

OAC

Organ Acquisition Costs

UNOS

United Network of Organ Sharing

EVPI

Expected Value of Perfect Information

IOM

Institute of Medicine

CER

Comparative Effectiveness Research

OPO

Organ Procurement Organization