Cost of a quality-adjusted life year in liver transplantation: The influence of the indication and the model for end-stage liver disease score

Authors


Abstract

Cost issues in liver transplantation (LT) have received increasing attention, but the cost-utility is rarely calculated. We compared costs per quality-adjusted life year (QALY) from the time of placement on the LT waiting list to 1 year after transplantation for 252 LT patients and to 5 years after transplantation for 81 patients. We performed separate calculations for chronic liver disease (CLD), acute liver failure (ALF), and different Model for End-Stage Liver Disease (MELD) scores. For the estimation of QALYs, the health-related quality of life was measured with the 15D instrument. The median costs and QALYs after LT were €141,768 and 0.895 for 1 year and €177,618 and 3.960 for 5 years, respectively. The costs of the first year were 80% of the 5-year costs. The main cost during years 2 to 5 was immunosuppression drugs (59% of the annual costs). The cost/QALY ratio improved from €158,400/QALY at 1 year to €44,854/QALY at 5 years, and the ratio was more beneficial for CLD patients (€42,500/QALY) versus ALF patients (€63,957/QALY) and for patients with low MELD scores versus patients with high MELD scores. Although patients with CLD and MELD scores > 25 demonstrated markedly higher 5-year costs (€228,434) than patients with MELD scores < 15 (€169,541), the cost/QALY difference was less pronounced (€59,894/QALY and €41,769/QALY, respectively). The cost/QALY ratio for LT appears favorable, but it is dependent on the assessed time period and the severity of the liver disease. Liver Transpl 17:1333–1343, 2011. © 2011 AASLD.

The increasing pressure to improve the cost-effectiveness of medicine has not spared the field of liver transplantation (LT).1-7 A cost-effective intervention is generally defined as either (1) an intervention that achieves a benefit similar to that achieved by its alternative but at a lower cost or (2) an intervention that achieves a considerably greater benefit than that achieved by its alternative, even at a higher cost, as long as the additional cost is worthwhile.

The absence of an effective alternative to LT has restricted controlled studies; with no comparable therapy, cost-effectiveness analyses are considerably more complicated and must depend on numerous assumptions and estimations.2, 5 Therefore, 2 issues of health economics here are the actual costs of LT with respect to the achieved health benefits [preferably measured as quality-adjusted life years (QALYs)] and the justification of these costs.2, 7, 8

LT is known to be very costly, but studies that quantify the cost with respect to QALYs are scarce.9-11 Similarly, although factors have been identified that can predict the high consumption of economic resources,3, 6, 12-19 the effects of these factors on the cost-effectiveness ratio are largely unknown. For instance, the high consumption of resources due to the allocation of organs to the sickest patients first might be balanced by large gains in the quality of life; however, cost-effectiveness ratios with respect to Model for End-Stage Liver Disease (MELD) scores remain uncalculated. Existing studies that have addressed economic issues often have included only certain LT indication groups, and they have commonly excluded patients with acute liver failure (ALF); moreover, the types of costs included in the calculations differ between the studies.3, 6, 9, 10, 12, 14, 15, 20 In addition, the focus is often confined to the immediate posttransplant period; thus, relevant long-term expenses, including outpatient follow-up and medications, are left unconsidered.3, 6, 12-15 In fact, most costs are incurred close to the time of the transplant operation, but survival and quality-of-life gains extend well beyond that period; thus, LT-related cost-effectiveness analyses should cover sufficiently long time periods.

Studies that take these issues into consideration could provide an important perspective for health policy discussions. They could also help with optimizing the tradeoff between the utilization of resources and the quality of care.

This study was designed to quantify the cost per QALY in a Finnish LT population with up to 5 years of follow-up. Cost-effectiveness was assessed separately for patients with chronic liver disease (CLD) and ALF and in various MELD score groups.

Abbreviations:

15D, 15-dimensional; ALF, acute liver failure; CI, confidence interval; CLD, chronic liver disease; HCC, hepatocellular carcinoma; HRQOL, health-related quality of life; LT, liver transplantation; MARS, Molecular Adsorbents Recirculating System; MELD, Model for End-Stage Liver Disease; NS, not significant; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; QALY, quality-adjusted life year.

PATIENTS AND METHODS

In Finland, all LT procedures are performed at 1 center, the Helsinki University Central Hospital, which is the largest of 5 university hospitals in the country. Patients are treated at the transplant center for all medical care–related issues; this treatment begins when patients are placed on the waiting list for LT and extends to approximately 1 year after LT. After that, the responsibility for primary care is shifted to the patients' residential hospital districts, but they also return to the transplant center regularly for follow-up visits and for hospitalization (when this is necessary).

The main study group comprised all adult patients who underwent LT for the first time between January 1, 2000 and December 31, 2006 in Finland. Each patient was followed until December 31, 2007 or death. All liver grafts were from brain-dead donors.

The clinical and demographic data came from the Finnish LT registry and from patient records. Pretransplant MELD scores were calculated at the time of listing for LT. No diagnosis-based exception points were added to the MELD scores. When the plasma creatinine level exceeded 350 μmol/L, a creatinine level of 350 μmol/L was assumed for the MELD equation. Some patients were receiving dialysis or treatment with the Molecular Adsorbents Recirculating System (MARS) at the time of their placement on the waiting list for LT. For those patients, the MELD score was calculated with the creatinine and bilirubin levels that had been recorded before the treatment was initiated, and the same upper limit of 350 μmol/L was used for creatinine.

Patients with CLD were divided into 3 clinically meaningful subgroups based on MELD scores: (1) MELD scores < 15, (2) MELD scores of 15 to 25, and (3) MELD scores > 25. Patients with ALF were divided into 2 subgroups based on the median MELD score at our center: MELD scores ≤ 30 and MELD scores > 30.

Cost Data

The direct costs and the survival rates were recorded for the main study group. The hospital costs were obtained from the Ecomed clinical patient administration database (Datawell, Ltd., Espoo, Finland); this is the hospital's official cost management system in which all costs are routinely stored. The costs included inpatient stays (ward and intensive care unit), outpatient visits, transplant operations, organ procurement, radiology, pathology, laboratory tests, blood products, staff costs, and pharmaceuticals (including in-hospital immunosuppressive drugs) associated with the Hospital District of Helsinki and Uusimaa. The 1-year cost data were assessed for the entire main study group and included costs from the time of placement on the waiting list for LT; for patients who were listed during a hospitalization, the cost data started from the beginning of that hospitalization episode. Thus, the costs of the LT evaluation process were captured for patients with ALF (who were always hospitalized at the time of listing) and for patients with CLD who required hospitalization. Furthermore, complete secondary care costs incurred more than 1 year after transplantation were available for a subset of patients who resided within the Hospital District of Helsinki and Uusimaa [81 patients (32% of all patients)].

The costs of medications (other than those issued during hospitalization) from the wait-list date to the end of follow-up were obtained from the Social Insurance Institution of Finland, which maintains complete records of all reimbursed drug purchases in the country. Medication costs were separated into 2 groups: (1) immunosuppression-related drugs and (2) other drug agents. Cost data on immunosuppressive drugs were obtained for each year. Costs of nonimmunosuppressive drugs were obtained for the entire follow-up period and then were divided by the years of follow-up for each patient so that the annual cost could be calculated.

All costs were inflated to the 2010 price level and were discounted at a rate of 3% in line with Finnish guidelines.21

Health-Related Quality of Life (HRQOL) Instrument

The effectiveness of LT was assessed in terms of changes in HRQOL measured before and after LT with the 15-dimensional (15D) instrument.22 This utility-based, validated, generic, self-administered questionnaire has 15 dimensions: moving, seeing, hearing, breathing, sleeping, eating, speech, eliminating (urination and defecation), everyday activities (keeping up with work, studies, household activities, and leisure activities), mental function, discomfort and symptoms, depression, distress, vitality, and sexual activity.22 Preference weights obtained from the general public through a 3-stage valuation procedure generate the utility score (the 15D score) over all the dimensions on a 0 to 1 scale [(0) deceased and (1) no problems in any dimension]. In most of the important properties, the 15D instrument compares favorably with similar instruments.23-27

HRQOL Before LT

Longitudinal HRQOL data were unavailable for the main study group. Thus, we used HRQOL data that were prospectively recorded at listing and 1 year after LT for patients placed on the waiting list in 2007 or later. The median 15D scores of the CLD patients were calculated for each MELD score–based subgroup. Patients with ALF before LT were not generally in a condition that allowed the completion of questionnaires of this kind. Thus, for patients with ALF undergoing LT between 2007 and 2010, the pretransplant 15D scores were estimated retrospectively by an experienced intensive care unit doctor (A.-M.K.) and an experienced transplant surgeon (H.I.) on the basis of patient records.

In addition, patients with grade 4 hepatic encephalopathy before LT, regardless of their indications, were assigned a 15D score of 0.0162, which was previously established for unconscious patients.22

HRQOL After LT

For those who were placed on the waiting list in 2007-2010 and responded to the 15D questionnaire 1 year after LT, we calculated median MELD group–specific 15D scores to depict the posttransplant HRQOL levels. On the basis of our previous findings28 and observations during annual follow-up visits at the transplant center, patient HRQOL levels were assumed to be maintained at the 1-year level throughout years 2 to 5. During the first posttransplant year, HRQOL was assumed to improve linearly from the level before LT to that at 1 year after LT. The HRQOL level of deceased patients was set to zero at the time of death, but it was assumed to decline to zero linearly during the last year of survival.

Economic Evaluation

The cost-utility analysis was performed from the perspective of the provider of secondary health care, and 2 time periods were used: from placement on the waiting list to 1 year after LT and from placement on the waiting list to 5 years after LT. The cost/QALY ratio was calculated separately for CLD, ALF, and MELD score–based subgroups.

A 1-way sensitivity analysis was conducted to determine how variations in the discount rate, costs, HRQOL, and survival affected the cost/QALY ratio. A worst-case analysis was conducted with the upper 95% confidence interval (CI) limits for costs, with the lower 95% CI limits for survival, and with the 25% quartiles for pre-LT and post-LT HRQOL.

Statistical Methods

Analyses were performed with PASW Statistics 18.0.0 (SPSS, Inc., Chicago, IL). The Mann-Whitney test was used for comparisons of 2 groups, the Kruskall-Wallis test was used for comparisons of 3 groups, the chi-square test was used for categorical variables, and the log-rank test was used for survival. P values < 0.05 were considered statistically significant.

Ethical Considerations

Ethical approval for the study was obtained from the ethics committee of the Hospital District of Helsinki and Uusimaa (registration number 538/E0/02). All patients signed an informed consent form before they completed the 15D questionnaire.

RESULTS

Patient Characteristics

The main study group comprised 252 patients (mean follow-up = 4.0 ± 2.1 years; Table 1). HRQOL data were obtained from a separate cohort of 78 patients (the HRQOL group). These 2 groups were similar with respect to the relevant clinical characteristics (Table 1).

Table 1. Characteristics of the Patients in the Main Study Group and in the Group Tested for HRQOL
 Main Study Group (n = 252)HRQOL Group (n = 78)P Value
  • *

    The data are presented as means and standard deviations.

  • Nine had viral cirrhosis.

  • Two had viral cirrhosis.

Age at transplantation (years)*49 ± 1248 ± 13NS
Sex [n (%)]  NS
 Male135 (54)34 (44) 
 Female117 (46)44 (56) 
Indication [n (%)]  NS
 CLD200 (79)64 (82) 
  PSC53 (21)16 (21) 
  PBC27 (11)8 (10) 
  Alcoholic cirrhosis36 (14)12 (15) 
  Liver tumor18 (7)6 (8) 
  Other66 (26)22 (28) 
 ALF52 (21)14 (18) 
  Drug-related9 (4)5 (6) 
  Budd-Chiari syndrome4 (2)0 (0) 
  Other10 (5)2 (3) 
  Unknown29 (11)7 (9) 
Pretransplant dialysis or MARS [n (%)]62 (25)15 (19)NS
Waiting list (days)*41 ± 6252 ± 64NS
 CLD50 ± 6663 ± 66NS
 ALF5 ± 82 ± 2NS
Retransplantation [n (%)]11 (4)4 (5)NS
CLD [n (%)]20064NS
 MELD score < 15120 (60)43 (67) 
 MELD score = 15-2560 (30)14 (22) 
 MELD score > 2520 (10)7 (11) 
ALF [n (%)]5214NS
 MELD score ≤ 3025 (48)5 (36) 
 MELD score > 3027 (52)9 (64) 

Two hundred of the 252 patients (79%) underwent LT for CLD, and 52 (21%) underwent LT for ALF. The average length of hospital stays during the first year, including protocol visits for patients traveling a long distance to the transplant center, was 43 ± 27 days for all patients, 41 ± 27 days for patients with CLD, and 52 ± 25 days for patients with ALF.

In comparison with patients with CLD, patients with ALF were significantly more likely to be female (P = 0.002) and had higher rates of pretransplant dialysis or MARS treatment (P < 0.001); these rates also increased with the MELD score in the CLD and ALF groups (Table 2).

Table 2. Clinical Characteristics of the Main Study Group According to the Indications and the MELD Scores
 CLD PatientsALF PatientsP Value
MELD Score < 15MELD Score = 15-25MELD Score > 25P Value*MELD Score ≤ 30MELD Score > 30P Value
  • *

    Significant differences between the MELD groups with CLD are shown.

  • Significant differences between the MELD groups with ALF are shown.

  • Significant differences between the CLD and ALF groups are shown.

Mean age at transplantation (years)495151NS4845NSNS
Female sex (%)443835NS6070NS0.002
Retransplantation rate (%)425NS124NSNS
Pretransplant MARS or dialysis (%)01765<0.00160890.016<0.001
3 main indications (%)PSC (39), PBC (17), and HCC (11)Alcoholic cirrhosis (33), cryptogenic cirrhosis (15), and PBC (12)Alcoholic cirrhosis (30), cryptogenic cirrhosis (20), and hepatitis C cirrhosis (15)<0.001Unknown (60), autoimmune hepatitis (20), and Budd-Chiari syndrome (12)Unknown (52), drug-related (30), and Budd-Chiari syndrome (4)NS<0.001

The predominant underlying disease for CLD patients with a MELD score < 15 was primary sclerosing cholangitis (PSC; many had premalignant findings in the bile ducts); in groups with higher MELD scores, alcoholic cirrhosis and cryptogenic cirrhosis predominated (Table 2). Among patients with ALF, unknown disease etiologies were accompanied predominantly by autoimmune hepatitis in the low–MELD score group and by drug-related causes in the high–MELD score group (Table 2).

Costs

The median total 1-year costs were €141,768 (Table 3). The median costs for the second posttransplant year were €9678, the median annual costs decreased thereafter by approximately €800 each year (Fig. 1), and the median total costs from listing to 5 years after LT added up to €177,618 (Table 3). Hospital costs during the first year (including the costs of the transplant procedure) were 75% (€132,951) of these total 5-year costs (Table 3). After the first year, the main cost was immunosuppression drugs, which were responsible for 59% of the annual costs (Table 3).

Table 3. Specifications for the Costs From Listing to 5 Years After LT for the Main Study Group (2000-2007)
  All PatientsCLD PatientsALF Patients
Listing to 1 YearAnnual Costs for Years 2 to 5Listing to 1 YearAnnual Costs for Years 2 to 5Listing to 1 YearAnnual Costs for Years 2 to 5
  • NOTE: Costs are presented in euros inflated to the 2010 level (according to statistics on purchasing power parities and exchange rates from the Organisation for Economic Co-Operation and Development, 1 euro is equal to 1.38919 US dollars).

  • *

    This includes the costs of all outpatient treatments (including dialysis and cancer treatments administered on an outpatient basis).

  • This includes the costs of inpatient stays (the ward and the intensive care unit), radiology, pathology, laboratory tests, blood products, and pharmaceuticals (including immunosuppressive drugs administered in the hospital).

Total costs (€)Median141,7689581132,7169772188,7397121
Minimum/maximum83,226/577,232 83,226/577,232 94,162/443,833 
25%/75% quartiles119,828/173,4256517/13,582117,437/159,4616652/13,861150,242/239,7124164/15,160
All hospital costs (€)Median132,9512250123,0582296169,0571854
Minimum/maximum73,401/576,993 73,401/576,993 85,361/413,468 
25%/75% quartiles109,641/164,802852/4386107,745/151,798841/4489140,311/223,6901143/3199
 Transplant operationMedian67,725 69,326 71,936 
Minimum/maximum52,169/168,415 54,805/168,415 52,169/157,673 
25%/75% quartiles54,805/75,618 54,805/75,142 57,478/77,483 
 Outpatient visits*Median196821991051830
Minimum/maximum0/34,465 0/34,462 0/9023 
25%/75% quartiles0/16520/10720/17090/11970/9330/626
 OtherMedian63,210115855,7091160101,6151064
Minimum/maximum1872/503,222 1872/503,222 33,191/320,216 
25%/75% quartiles46,159/94,641384/252643,770/79,102337/254576,875/150,899479/2410
Drugs (€)       
 Immunosuppressive  drugsMedian837956338428578276605276
Minimum/maximum0/27,656 0/25,666 0/27,656 
25%/75% quartiles5931/12,5493725/85226253/13,0043784/86954898/12,0413357/7752
 Other drugsMedian859616929686508386
Minimum/maximum0/16,864 0/16,864 0/3507 
25%/75% quartiles402/1426233/1107474/1595254/1319252/873106/687
Figure 1.

Mean costs from the time of placement on the waiting list for LT (listing) to 1 year after transplantation and annual costs up to 5 years after transplantation. The vertical lines represent standard deviations.

Ninety-five percent of all patients purchased nonimmunosuppressive drugs, and the median annual cost of these drugs was approximately €860. Furthermore, 94% of the patients purchased antimicrobial drugs, and 85% purchased antihypertensive drugs; the median annual costs were €30 and €98, respectively. Approximately one-third of the patients purchased antidiabetes drugs and lipid-lowering drugs. Nonimmunosuppressive drugs other than those previously mentioned were purchased by 95% of the patients with a median annual cost of €420.

The median total cost for patients with ALF from placement on the waiting list to 1 year after LT (€188,739) was 42% greater than the cost for patients with CLD (€132,716; Table 3). This difference resulted primarily from the marked differences in non-LT hospital costs (including intensive care and MARS treatments). However, the median annual costs for years 2 to 5 were lower for patients with ALF (€7121) versus patients with CLD (€9772). The median cost of immunosuppressive drugs was slightly higher in the CLD group versus the ALF group (Table 3).

In the CLD and ALF groups, the median total costs from listing to 1 year after LT were higher for patients with high MELD scores versus those with low MELD scores (Table 4). Years 2 to 5 showed the opposite trend; the median annual costs were €10,927 for CLD patients with MELD scores < 15, €7872 for CLD patients with MELD scores of 15 to 25, €7752 for CLD patients with MELD scores > 25, €7388 for ALF patients with MELD scores ≤ 30, and €6271 for ALF patients with MELD scores > 30.

Table 4. Costs, Survival Rates, QALYs, and Costs per QALY From Listing to 1 and 5 Years After LT in the Main Study Group (n = 252)
 CostsSurvival (%)HRQOLMedian QALYsMedian Cost per QALY (€)
Listing to 1 YearListing to 5 YearsBefore LTAfter LT
nMedian (€)n*Median (€)At 1 YearAt 5 YearsnMedianUnconscious Patients (%)Adjusted MediannMedianAt 1 YearAt 5 YearsAt 1 YearAt 5 Years
  • NOTE: The transplant period was 2000-2006; QALYs were estimated from the quality-of-life data for a separate cohort undergoing transplantation in 2007-2010 (n = 78). Costs are presented in euros inflated to the 2010 level (according to statistics on purchasing power parities and exchange rates from the Organisation for Economic Co-Operation and Development, 1 euro is equal to 1.38919 US dollars). The 15D score at 1 year was extrapolated to the following years.

  • *

    Patients resided within the Hospital District of Helsinki and Uusimaa.

  • Patients with grade 4 hepatic encephalopathy in the main study population (they were assigned the standard 15D score of 0.0162 for unconsciousness).

  • Adjusted for the proportion of unconscious patients.

All patients252141,76881177,6189588780.7887.10.765470.8910.8953.960158,40044,854
CLD patients200132,71671171,5749688640.7993.50.798380.8980.9274.037138,68142,500
 MELD score <15120132,46040169,5419891430.8261.70.822290.9010.9484.059139,72641,769
 MELD score = 15-2560137,90820163,5399786140.8228.30.78660.9060.9124.159151,21539,322
 MELD score > 2520184,41611228,434787570.64000.64030.8750.7723.814238,88159,894
ALF patients52188,73910211,4118988140.23621.00.21890.8680.5403.306349,51763,957
 MELD score ≤ 3025187,6915211,702929150.45720.00.34140.9690.7164.036262,13852,453
 MELD score > 3027191,0415211,120858490.21822.20.20850.8590.5383.474355,09560,771

The 5-year costs were markedly lower for patients who underwent LT once (median = €175,028, n = 241) versus those who underwent retransplantation once (median = €324,298, n = 10) or twice (median = €630,444, n = 1).

Patient Survival

The overall patient survival rate was 95% at 1 year and 88% at 5 years (Table 4). Significant differences in the survival rates emerged between the CLD MELD groups (P = 0.04) but not between the ALF MELD groups or between the CLD and ALF groups.

HRQOL

After adjustments for the proportions of patients with grade 4 hepatic encephalopathy (3.5% of the CLD patients and 21.0% of the ALF patients), the median 15D score before LT was lower for patients with ALF (0.218) versus patients with CLD (0.798, P < 0.001; Table 4). Significant differences also emerged between the CLD MELD groups (P = 0.03) but not between the ALF MELD groups (Table 4). The posttransplant HRQOL levels differed nonsignificantly between the indication groups and the MELD groups.

The posttransplant gain in HRQOL versus the pretransplant level was substantially larger for the ALF patients versus the CLD patients (Table 4). For both ALF and CLD patients, the gain was somewhat higher in groups with higher MELD scores (Table 4).

Economic Evaluation

The 1-year median number of QALYs for all patients was 0.895, and the cost was €158,400/QALY (Table 4). The respective figures were 0.927 and €138,681/QALY for the CLD group and 0.540 and €349,517/QALY for the ALF group (Table 4). The 5-year cost decreased to €44,854/QALY for all patients, to €42,500/QALY for CLD patients, and to €63,957/QALY for ALF patients (Table 4). Table 4 shows the cost per QALY by MELD groups.

Sensitivity Analysis

The ALF group outcomes (particularly survival) were more sensitive than the CLD group outcomes to variations in input values (Table 5). The worst-case scenario increased the 5-year cost per QALY by 40% in CLD patients and nearly 2-fold in ALF patients (Table 5).

Table 5. One-Way Sensitivity Analysis Showing the Impact of Variations in Input Values on the Median Cost/QALY Ratio at 5 Years
 Costs (€)
All PatientsCLD PatientsALF Patients
  • NOTE: Costs are presented in euros inflated to the 2010 level (according to statistics on purchasing power parities and exchange rates from the Organisation for Economic Co-Operation and Development, 1 euro is equal to 1.38919 US dollars).

  • *

    This assumes a linear HRQOL improvement to the posttransplant level within 6 months after LT.

  • This assumes that HRQOL is maintained at the pretransplant level for 6 months after LT, then improves linearly, and reaches the posttransplant level within 1 year after LT.

  • The lower CI or quartile limits for pretreatment and posttreatment HRQOL and survival and the upper limits for costs.

Base case (discounted at 3%)44,85442,50063,957
Discount rate (0%)41,80939,70257,578
Total costs   
 Upper 95% CI limit47,20444,92770,743
 Lower 95% CI limit42,50240,07457,152
Pretreatment HRQOL   
 25% quartile50,23043,94568,149
 75% quartile41,88542,10959,727
Posttreatment HRQOL   
 25% quartile47,33145,08366,369
 75% quartile42,99041,00258,327
HRQOL improvement after transplantation   
 Rapid*44,62742,29560,955
 Delayed45,06642,71967,206
Survival   
 Upper 95% CI limit42,64338,70049,440
 Lower 95% CI limit47,12248,99781,143
Worst-case scenario54,57959,536133,434

DISCUSSION

As a lifesaving, effective, and occasionally urgently required intervention, LT is justified in modern medicine on various ethical grounds that extend beyond cost-utility issues.29-32 Nonetheless, with the demand for health care threatening to exceed the increase in resources, an expensive therapy such as LT deserves a careful economic evaluation and efforts to improve its efficiency.

The present study weighed the cost of LT against the QALYs achieved and investigated how LT indications and MELD scores influenced that ratio. The median total 5-year cost of LT amounted to €177,618. Because the 1-year cost was 80% of the total costs, the cost/QALY ratio improved yearly from €158,400/QALY at 1 year to €44,854/QALY at 5 years. The annual costs after the first posttransplant year were relatively low, and although these costs continued to appear, the number of QALYs also grew. Hence, because our results only extended over 5 years, we may have underestimated the true cost-effectiveness of LT.

The cost/QALY ratio was more favorable for CLD patients versus ALF patients and for patients with low MELD scores versus patients with high MELD scores. The overall 5-year cost/QALY ratio remained rather stable, despite variations in the sensitivity analysis.

The principal strengths of this study include the relatively long follow-up of a cohort with no a priori patient exclusions and the incorporation of data from patients who were unconscious and severely ill before LT (such patients are typically unable to complete HRQOL questionnaires). Moreover, we accurately captured comprehensive cost data, which included hospitalization costs, costs incurred during follow-up, and drug costs.

A few major drawbacks merit our attention. First, the cost data and the HRQOL data came from different patient cohorts. Also, the pretransplant HRQOL levels of the ALF group were based on retrospective expert estimations. Reliability and applicability were, however, achieved through the linking of the indications and the MELD scores. The 2 populations were also similar in all relevant clinical characteristics; nevertheless, unmeasured differences may have existed.

Second, the absence of a realistic alternative to LT placed restrictions on the calculations of incremental cost-utility ratios. Although survival without LT (shadow survival) can be estimated by the MELD score for many indications, no methods exist for estimating shadow HRQOL levels and shadow costs for different indications and MELD scores. Cost data for historical conservative therapies or for patients with contraindications for LT are not precisely applicable to modern LT populations.5 Without reliable shadow data, an implicit comparison is made to immediate death, which has no costs or QALYs. Under this assumption, our cost/QALY results could be interpreted as incremental cost-utility ratios. In reality, however, in the absence of LT, patients would accrue some costs and QALYs before death (particularly CLD patients with low MELD scores).

Third, the costs incurred by patients on the waiting list who died before LT could not be captured. The wait-list mortality rate in Finland is 2% to 5%.

Fourth, indirect costs (eg, missed work and the need for home health care) were not considered. Pretransplant evaluation costs were captured only for patients with CLD who were hospitalized when they were placed on the waiting list and for patients with ALF.

In Finland, organ allocation according to a specific score is of minor importance because 1 transplant center serves both recipients and donors, and the waiting times are relatively short (Table 1). Typically, only approximately 10 patients are on the waiting list at once. Thus, allocation is fairly comprehensive; after blood group and size matching, allocation proceeds according to the clinical conditions and waiting times of the patients. Candidates with ALF who require high-urgency LT have priority for any blood group–compatible donor organ within 72 hours in the Nordic collaboration called Scandiatransplant.

Direct comparisons of the costs of LT in this study with the costs in other studies are hampered by the different health care systems and the diverse eras of analysis. However, a recent meta-analysis of 30 studies3 estimated that the average inflation-corrected cost of LT [ie, the costs related to transplantation and the initial posttransplant hospital stay (mean stay = 1-2 months)] was $163,438 (95% CI = $145,277-181,598) in the United States and $103,548 (95% CI = $85,514-121,582) in other countries that are members of the Organisation for Economic Co-Operation and Development. We found that the 1-year hospital costs for the CLD group (€123,058) were similar to those of the meta-analysis, but the inclusion of retransplantation and a relatively high proportion of patients with ALF resulted in higher overall costs than those reported in the meta-analysis. We have reported results as medians instead of means because medians are less sensitive to extreme outliers (eg, costs from retransplantation) in small samples.

In agreement with other studies,12-14, 18, 19, 33 we found that ALF, a high MELD score, and retransplantation were cost drivers in the early term. The high costs of the ALF group originated mainly from the initial transplant admission. In the long term, the lower follow-up costs reduced the gap between the ALF and CLD groups; the 5-year costs of the CLD subgroup with the highest MELD scores exceeded the costs of the ALF group.

Among the CLD patients, the group with MELD scores > 25 may have had high early costs because of associated dialysis and MARS treatments, prolonged intensive care, and prolonged hospital stays. Among the ALF patients, the costs were quite similar between the MELD groups. Although the MELD score lacks robust validation for ALF patients, recent evidence has suggested that it is applicable in this setting.34-38 We used the MELD score mainly to enable comparisons between ALF and CLD.

A few previous studies that quantified the cost-utility of LT9-11 reported nonincremental cost/QALY ratios ranging from €22,225/QALY10 to £58,972/QALY9; these ratios depended on the methodology, the time and locale of the study, the assessed time period, and the included liver diseases. None of these studies, however, included ALF patients or evaluated the influence of disease severity on the cost/QALY ratio.

Currently, the MELD score is widely used as a marker of CLD severity for prioritizing patients for LT. For CLD patients, the MELD score also correlates with survival benefits from LT.39, 40 These favorable features of the MELD score are counteracted by concerns about higher costs, more complications, and lower survival rates early after LT for patients with high MELD scores.14, 19, 41, 42 In our study, the CLD group with MELD scores > 25 was associated with low QALYs and high cost/QALY ratios (in addition to high costs) in comparison with the CLD group with MELD scores ≤ 25. However, within 5 years after LT, the difference in the median cost/QALY ratios of the CLD groups with the highest and lowest MELD scores was narrowed to just €18,125. Because a high MELD score predicts markedly shorter survival without LT, this small cost/QALY difference should pose no significant conflict with the numerous competing ethical principles in favor of allocating organs to the sickest patients first.

The predominant disease etiologies varied among the MELD groups. Longworth et al.9 reported better cost-utility when LT was performed for PSC versus alcoholic cirrhosis. In our study, the CLD group with low MELD scores predominantly had PSC, whereas the CLD group with high MELD scores predominantly had alcoholic cirrhosis. Comparisons of the cost/QALY ratios of the high–MELD score and low–MELD score groups were also complicated by likely differences in shadow cost/QALY ratios, which were not captured here.

Interestingly, the costs were similar for the 2 MELD groups of ALF patients, but the cost/QALY ratio was better for those with MELD scores ≤ 30 versus those with MELD scores > 30.

Although cost/QALY analyses enable comparisons across a wide variety of medical interventions, these comparisons should primarily be performed within the context of similar medical circumstances; a therapy that prevents imminent death may not be comparable to one that only improves the quality of life.43 There are no benchmark figures that reflect society's willingness to pay for an intervention when the alternative is imminent death. However, some straightforward and immediately lifesaving treatments provided at our hospital have been studied with a similar methodology. For instance, the average direct medical cost/QALY ratio for MARS treatment in ALF patients was €64,732/QALY over 3.5 years44; the average direct medical cost/QALY ratio for general intensive care ranged from €38,405 to €118,668/QALY over 1 year and depended on the assumptions made (T. Vainiola, unpublished data, 2011). On the other hand, LT also compares favorably with home hemodialysis (€45,805/QALY) and self-care satellite dialysis (€46,801/QALY) for patients with chronic end-stage renal disease (a 1-year time frame with drug costs included).45

Findings from the Finnish health care system and the Finnish LT setting may be limited in their generalizability to other settings. For instance, the prevalence of hepatitis C in our LT population is low, the proportion of ALF patients is higher (21%) than that at most centers (∼10%), the waiting times are relatively short, and the Finnish population is relatively homogeneous in ethnographic and socioeconomic terms. Moreover, our findings are not directly applicable to settings in which living donors or donors after cardiac death are used.

In conclusion, LT is a lifesaving but costly therapy. We found that the majority of the costs were due to the initial transplant admission, and the cost/QALY ratio improved over time. A more beneficial cost/QALY ratio was observed for CLD patients versus ALF patients and for patients with low MELD scores versus patients with high MELD scores, but the gap between MELD groups was much narrower than the gap by costs alone. Whether the overall 5-year cost of €44,854/QALY allows LT to be considered a cost-effective intervention ultimately depends on the judge and his grounds for deciding.

Ancillary