The effectiveness and cost-effectiveness of methods of storing donated kidneys from deceased donors: a systematic review and economic evaluation

Incidence and prevalence

The Renal Registry annual report 2006 shows that there were 41,776 adults on RRT (see Management of end-stage kidney disease, later in this chapter) in the UK in 2005; this gives a prevalence of 694 per million population (pmp). There were also 748 children (< 18 years) on RRT with a prevalence of 12 pmp. These figures show that since the year 2000 there has been a 27.8% increase in patient numbers cared for by the 38 renal units which have continuously returned data from 2000 to 2005. 12

Data from the same report show that in 2005 there was an acceptance rate for RRT in the UK of 108 pmp for adults and 2 pmp for children, showing a total incidence of 110 pmp. This reveals a 7.3% increase in incidence from 2001 to 2005 in 42 renal units in the UK submitting full returns to the Renal Registry. 12 Figure 2 shows the incident rates for the UK from 1990 to 2005.

FIGURE 2.

Incident rates of adults accepted for renal replacement therapy in the UK 1990–2005. (Source: The data reported here have been supplied by the UK Renal Registry of the Renal Association. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the UK Renal Registry or the Renal Association.)

In 2005 in the UK, 76% of people accepted for RRT began treatment with haemodialysis (HD), 21% started with peritoneal dialysis (PD) and 3% with a kidney transplant. Ninety days later 8% had died and 1% had stopped treatment or had been transferred out. Of the remaining 91%, 5% changed from HD to PD and 3.2% had a transplant. 12 The median age at which people start RRT has increased in England from 63.8 years in 1998 to 65.2 years in 2005, with people using HD having a mean age of 9 years older and having fewer co-morbidities than those using PD. 12 Table 1 shows the percentage RRT type for new patients in England and Wales in 2005.

Survival in the first year following commencement of RRT for all patients regardless of age is 79%. 12 Five-year survival figures including deaths in the first 90 days following commencement of RRT are shown in Table 2.

Significance for patients

To a person suffering from ESRD, the opportunity to have a kidney transplant is literally a matter of life or death. In the year 2006–7, in the UK, 231 patients died while on the active and suspended waiting lists for kidney transplantation; an equivalent number were removed from the list because they were no longer fit enough, most of whom went on to die. In the same year there was an 11% increase in patients actively waiting for a kidney or kidney and pancreas transplant compared with the previous year, with a total of 6480 people waiting for a transplant. Seventeen per cent (1101) of those on the 2006–7 waiting list had received a transplant by 31 March 2007. 18 Figure 4 shows the percentage of dialysis patients who survived in 2005.

FIGURE 4.

One-year UK survival of prevalent dialysis patients in different age groups 2005. Lines show 95% confidence intervals. (Source: The data reported here have been supplied by the UK Renal Registry of the Renal Association. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the UK Renal Registry or the Renal Association.)

Quality of life

Cold storage

In cold static storage, the kidney is flushed through with a preservation solution and kept on ice. Two preservation solutions are widely used in the NHS for cold storage: Marshall’s hypertonic citrate (Soltran™, Baxter Healthcare) and University of Wisconsin (ViaSpan™, Bristol Myers Squibb). Cold storage solutions used in other health systems are: Celsior™ (Genzyme), Histidine–tryptophan–ketoglutarate (HTK, Custodiol) and Euro Collins (Fresenius). The characteristics of these solutions can be seen in Table 5. Preservation solutions used in cold static storage are different from those used in machine perfusion.

Three cold storage solutions will be considered in this assessment. These are Viaspan, Soltran and Celsior. The first two have been selected because they are in current NHS use; additionally, Celsior will be included because it has been relatively newly developed and may become used in the UK.

The other cold storage solutions will not be considered because they are outside the scope of this assessment.

The benefits of simple cold storage are that it is not labour intensive, organ exchange is easy and there are no additional risks of damaging the kidney.

Hypothermic machine perfusion

In hypothermic machine perfusion, core cooling of the kidney is maintained by continuously pumping cold preservation solution through it. This solution also provides nutrients, sometimes oxygen, carries away toxic metabolites and provides ‘buffering’ (reducing build up of lactic acid). In theory, this process should reduce the damage associated with CIT. Machine perfusion can be used to preserve grafts from both BSD and DCD donors. However, in the UK they are predominantly used for DCD donors or kidneys with an anticipated long ischaemic time. It is suggested that assessments carried out during machine perfusion may also provide information about the viability of kidneys for transplantation which would aid the selection of grafts. 45 Up to 10% of kidneys from DCD donors never function after transplantation, predominantly those from uncontrolled donors. 9

The disadvantages of machine perfusion are that it is more labour intensive, is less practical in organ exchange and potentially risks damage to the renal artery.

Two commercially available machine perfusion systems have been identified. One is the LifePort Kidney Transporter^® (Organ Recovery Systems), a portable system which can perfuse one kidney and can run without being overseen. The other machine is the RM3 Renal Preservation System^® (Waters Medical Systems); this non-portable system can perfuse two kidneys simultaneously but needs to have its running supervised. It is not intended to be transportable between hospitals and is not used in the UK. A perfusion solution with a formula developed at the University of Wisconsin is used with machine perfusion (sometimes known as University of Wisconsin machine preservation solution or Belzer MPS; it is sold under the brand name KPS-1 by Organ Recovery Systems for use with their machine).

Two other hypothermic perfusion machines have been identified in development; these are TRANSren™ (Organ Assist, www.organ-assist.nl) and Airdrive™ (Indes, www.indes.eu). TRANSren research has only taken place in animals; similarly the Airdrive disposable perfusion system has only had research conducted in animals and in the human liver. Therefore, owing to the lack of comparative human kidney studies, these devices will not be included in this assessment.

Study design

Interventions and comparators

Each intervention was compared with all the others, data permitting.

Two methods of cold storing kidneys were considered: hypothermic machine perfusion and cold static storage solutions. Both these technologies were reviewed from the perspective of the UK NHS and so we only considered those specific products that are either in current use or are likely to be available and comparable with those currently used. We did not look at studies of kidney storage technologies that predate current technologies and have been shown to be technically inferior or are not available in the UK.

Machine perfusion interventions considered were:

• LifePort Kidney Transporter (Organ Recovery Systems)

• RM3 Kidney Preservation System (Waters Medical Systems).

Cold storage solutions considered were:

• University of Wisconsin (ViaSpan, Bristol Myers Squibb)

• Marshall’s (Soltran, Baxter Healthcare)

• Celsior (Genzyme).

For more details of the processes of machine perfusion and cold storage see Description of technology under assessment, Chapter 1.

Population

The population assessed are recipients of transplanted kidneys from deceased donors. These can be:

• BSD: death is diagnosed by absence of any brain stem activity, although the heart is still beating.

• DCD: death is diagnosed by cessation of the heart beat. These can be further subdivided into those whose cardiac arrest occurred in a controlled or in an uncontrolled setting.

• ECDs: donors who are either over 60, or are over 50 and with at least two of the following features: a history of hypertension, death by a cerebral vascular accident or terminal creatinine levels > 1.5 mg/dl.

More details of the characteristics of the population can be found in Epidemiology, Chapter 1.

Outcomes

The outcomes of interest include:

• discard rates of non-viable kidneys post storage

• incidence DGF

• incidence of PNF

• patient survival

• graft survival

• graft rejection rates

• graft function measured by creatinine concentrations and glomerular filtration rate

• adverse events.

These outcomes are more fully described in Chapter 2.

Internal validity

Consideration of internal validity addressed:

1. Sample size

   1. power calculation at design – for RCTs

2. Selection bias

   1. explicit eligibility criteria

   2. proper randomisation and allocation concealment – for RCTs

   3. similarity of groups at baseline

3. Performance bias

   1. similarity of treatment other than the intervention across groups

4. Attrition bias and intention-to-treat (ITT) analysis:

   1. all kidneys are accounted for

   2. number of withdrawals specified and reasons described

   3. analysis undertaken on an ITT basis

5. Detection bias

   1. blinding

   2. objective outcome measures

6. Appropriate data analysis.

External validity

External validity was judged according to the ability of a reader to consider the applicability of findings to a patient group and service setting. Study findings can only be generalisable if they describe a cohort that is representative of the affected population at large. Studies that appeared representative of the UK kidney transplant population with regard to these considerations were judged to be externally valid.

Number of studies included

One hundred and thirty-six full papers were reviewed to see if they met the inclusion criteria. In addition, ongoing studies were considered. Thirteen articles were found that met the inclusion criteria, leaving 123 exclusions. A flow chart of papers through the review process (Figure 43) including reasons for exclusion can be found in Appendix 2, and a list of studies excluded at the paper review stage is given in Appendix 4.

The 13 articles included were: two systematic reviews,47,49 three full journal published RCTs,50–52 two RCTs,53,54 one cohort study,55 three full journal published retrospective record reviews56–58 and two retrospective record reviews published as posters and abstracts only. 59,60

Further examination of the systematic reviews showed that the review conducted by Wight and colleagues (2003)47 did not include any studies that met the inclusion criteria for this systematic review, as at least one comparator in every study was of an older technology and outside the scope of this report. Therefore, this systematic review was excluded.

The other systematic review, by Costa and colleagues (2007),49 updated Wight and colleagues. They found 10 new studies, one of which,61 seemed to meet our inclusion criteria. However, upon further examination it was found that there was not sufficient information for critical appraisal; the authors were contacted but little further information was gleaned. Therefore, this study and the systematic review it came from were excluded. See Table 8 for a comparison of study type and publication status.

Upon further examination of the papers it emerged that in one of the trials54 cold storage using both ViaSpan and HTK cold storage solutions was allowed. However, the data were not disaggregated, making analysis of the ViaSpan results alone impossible. We therefore conducted further searches for studies comparing HTK with our interventions and found 10 studies. One of these was an RCT comparing ViaSpan and HTK. 63 This showed that the solutions were broadly equivalent in terms of kidney graft and patient outcomes with BSD donated kidneys. The other papers found did not fill in any evidence gaps in our study comparisons table, so we decided to exclude them, but to allow papers that used a combination of ViaSpan and HTK for cold storage, as we considered these to be comparable. Table 9 shows a matrix of the comparisons of interest in this assessment; shaded cells illustrate which comparators were investigated.

No calculations were made to assess the level of agreement on selection or validity decisions.

Summary of included studies’ characteristics

Table 10 contains a summary of the key design characteristics of the included studies. Data extraction tables for each study can be found in Appendix 3.

A summary of assessment of the quality of our included studies can be found in Table 11.

LifePort versus ViaSpan

Of the three studies comparing LifePort with ViaSpan, one is an ongoing RCT (Watson and colleagues),53 one RCT has not completed economic data analysis (Moers and colleagues)54 and the other is a retrospective review of hospital records. 56

Watson and colleagues53 (academic-in-confidence information removed). Moers and colleagues54 [Germany, the Netherlands and Belgium, n = 1086 (kidneys)] conducted a good quality European multicentre RCT [the Machine Preservation Trial (MPT)]. This study randomised 1086 kidneys from DCD and BSD (Maastricht criteria III and IV) donors to LifePort (n = 543) or ViaSpan or HTK (n = 543). Immediately post randomisation, 368 kidneys were excluded. Randomisation allocation was kept for 1036 kidneys and broken for 50; this was permitted only when the anatomy of the kidney made machine perfusion unsuitable. Subsequently, 42 kidneys were discarded post storage and prior to transplant for a variety of reasons (if a kidney was excluded from one arm then its contralateral pair was excluded from the other arm) [LifePort = 11 (+10), ViaSpan = 10 (+11)] and four were excluded post transplant [LifePort = 0 (+2), ViaSpan = 2]. This left 672 kidneys for data analysis (BSD = 588, DCD = 84): LifePort, n = 336; ViaSpan, n = 336. In total, 414 kidneys were excluded post randomisation. Recipients who died in the first week post transplant were excluded from the analysis. See Appendix 5 for details of reasons for exclusions.

Recipients were children and adults, mean age 52.5 years (range 2–79 years) who were followed up for 1 year. There were no significant differences in baseline characteristics, including CIT (mean 15 hours) (ViaSpan = 2.5–29.7, LifePort = 3.5–29.7). Results were reported at 6 and 12 months. Analysis was not by ITT. However, in the context of a trial with matched kidneys, the lack of ITT analysis might be considered less of a threat to internal validity.

The results showed that for the primary end point of DGF there were no significant differences between the storage methods (LifePort = 70, ViaSpan = 89, p = 0.05).

The secondary end point of duration of DGF showed that machine perfusion significantly reduced this parameter [LifePort = 10 (1–48), ViaSpan = 13 (1–41), p = 0.04]. Another measure, functional DGF [an absence of a decrease in serum creatinine of at least 10% per day for at least 3 consecutive days in the first week after transplant, not including patients who developed acute rejection and/or calcineurin inhibitor (CNI) toxicity in the first week] was added post hoc and not specified in the trial protocol; this outcome showed a reduced incidence for machine perfusion (LifePort = 77, ViaSpan = 101, p = 0.03). Other secondary outcome measures showed no significant differences between the two groups (PNF, acute rejection, creatinine clearance at day 14, CNI toxicity within 14 days and post-transplant hospital stay).

At 6 months post transplant, there were no differences in patient survival between the groups; both were 98% [relative risk (RR) 1.00, 95% confidence interval (CI) 0.98 to 1.00]. Similarly, at 12 months post-transplant, patient survival was 97% for both groups (RR 1.00, 95% CI 0.97 to 1.03).

Graft survival at 6 months failed to show a significant difference between the groups. However, the LifePort group had significantly better graft survival at 12 months post transplant [LifePort = 329/336, ViaSpan = 316/336: hazard ratio (HR) 0.39, 95% CI 0.17 to 0.89, p = 0.03).

When the results were censored for death at 12 months, Moers and colleagues found that for grafts that had been subject to DGF, those that had been machine perfused were more likely to survive (LifePort = 61/65, ViaSpan = 65/79: RR 1.14, 95% CI 1.01 to 1.29, p= 0.04). There were no significant differences for the death-censored survival of grafts that had not had DGF. Main results can be found in Table 12.

Moers and colleagues carried out subgroup analyses for DGF. In order to carry out this analysis, further DCD participants were enrolled (n = 80 or 82 donors – both numbers are given). They found no significant differences between standard criteria donors versus ECD, BSD versus DCD (main data set) or BSD versus DCD (extended data set).

Moustafellos and colleagues56 [UK, n = 36 (kidneys)] reviewed the previous 3 years’ records of patients receiving a DCD kidney (Maastricht class III or IV) at the Oxford Transplant Unit. They found that 18 people had received kidneys preserved by a LifePort machine and 18 by ViaSpan in cold storage. The two groups received different induction therapies and the mean age of the ViaSpan transplant recipients was older by 18 years (LifePort = 36 years, ViaSpan = 54.5 years, p < 0.001). The groups also varied in length of cold ischaemia [LifePort = mean 15 hours, ViaSpan = mean 17 hours; difference in means (DM) –1.5 hours, p < 0.001]. These differences in group characteristics and the potential for bias introduced by lack of randomisation mean that the results of this study must be interpreted with great caution.

Moustafellos and colleagues found that on their primary outcome measure of immediate renal function, kidneys stored by machine perfusion were more likely to work straightaway than those cold stored (LifePort = 13/18, ViaSpan = 2/18; RR 6.5, 95% CI 1.71 to 24.77, p < 0.001). Their secondary outcome measures were similarly significant: DGF (LifePort = 5/18, ViaSpan = 16/18; RR 0.31, 95% CI 0.15 to 0.67, p < 0.001); length of hospitalisation (LifePort = mean 8 days, ViaSpan = mean 14 days; DM –6, 95% CI –7.66 to –4.34, p < 0.001); and creatinine concentrations at discharge (DM –118 μmol/l, p < 0.001), all favouring machine perfusion.

Principal outcomes can be seen in Table 12.

LifePort versus Marshall’s cold storage solution (Soltran)

Plata-Munoz and colleagues55 [UK, n = 60 (kidneys)] conducted a sequential cohort study of DCD Maastricht category III controlled donor kidneys (March 2002–December 2005). For the first 2 years of the study, all kidneys were cold stored using Marshall’s solution (n = 30); subsequently, all kidneys were stored using the LifePort machine perfusion system (n = 30).

They found that the baseline characteristics of the groups were similar apart from mean recipient age [LifePort group = 47 years (range 20–69), Marshall’s = 54 years (range 34–76), p < 0.01]. Also, the mean CIT was slightly greater for kidneys stored by LifePort [LifePort group = mean 19 hours (range 15–23), Marshall’s = mean 18 hours (range 14–22)]. Clinical outcomes showed a lower proportion of DGF in the LifePort group (RR 0.64, 95% CI 0.43 to 0.89, p < 0.05) as well as length of hospital stay (LifePort = 10 days, Marshall’s = 14 days, p < 0.05). Graft function (serum creatinine) was better at 6 and 12 months for kidneys stored in the LifePort machine (6 months: DM –38.00 μmol/l, 95% CI –46.32 to –29.68, p < 0.001; and 12 months: DM –39.00 μmol/l, 95% CI –48.51 to –29.49, p < 0.001). Rates of acute rejection were low for both interventions [LifePort = 4/30 (13%), Marshall’s = 2/30 (7%)]. However, there were no significant differences between groups in patient or graft survival after 1 or 2 years. Results are presented in Table 13.

Summary of machine perfusion versus cold storage solutions

Four studies compared machine perfusion with cold storage; two were RCTs, one was a prospective cohort study and one was a hospital record review.

The donor populations for the two RCTs were different; with DCD donors in the Watson and colleagues trial and mostly BSD (with some DCD) donors in the Moers and colleagues study. The overall rate of DGF in the Moers and colleagues trial was a lot less than in Watson and colleagues [24% and (academic-in-confidence information removed) respectively]; this may have been due to the difference in DGF between DCD- and BSD-donated kidneys and the large numbers of kidneys that were excluded from Moers and colleagues post storage and prior to analysis (n = 42). However, Watson and colleagues found (academic-in-confidence information removed) and Moers and colleagues found less with LifePort (LifePort = 21%, ViaSpan = 26%).

Overall (academic-in-confidence information removed). However, Moers and colleagues found that graft survival was better at 12 months post transplant with machine perfusion (LifePort = 98%, ViaSpan = 94%, p = 0.03). Only 3 months’ follow-up data were available from Watson and colleagues, who found (academic-in-confidence information removed). These two studies’ results are in recipients whose grafts had a mean CIT of approximately 15 hours. It is not possible to say from these data what the effects of longer follow-up or greater CIT may have on the results.

Moers and colleagues carried out subgroup analyses for DGF. They found no significant differences between standard criteria donors (undefined) versus ECD, BSD versus DCD (main data set) or BSD versus DCD (extended data set).

In contrast, the results from the smaller cohort (Plata-Munoz and colleagues) and record review (Moustafellos and colleagues) studies found significant differences for DGF, length of hospital stay, and graft failure at 6 and 12 months favouring LifePort over ViaSpan and Marshall’s Soltran. Plata-Munoz and colleagues also reported patient graft survival outcomes at 1 and 2 years, but found no significant differences between groups. As these non-RCT results may have been influenced by selection bias and other confounding factors, they cannot be considered as internally valid as those from the two RCTs.

Where post-storage, pre-transplant discard rates were reported, these were similar between the two groups (academic-in-confidence information removed); MPT: machine perfusion = 11, cold storage = 10).

Abstracts and posters only

Summary of machine perfusion versus machine perfusion

We found only two studies assessing the comparative effectiveness of the LifePort and RM3 machine perfusion systems (Guarrera and colleagues and Kazimi and colleagues). These were both retrospective hospital record reviews that had not been through a peer-review process and had only been published as abstracts and presented as posters. Therefore, the evidence they present is unproven.

With the exception of PNF, all outcomes favoured the RM3 over the LifePort perfusion machine. Guarrera and colleagues found significant benefits for kidneys stored in the RM3 machine, for ECD and DCD donated kidneys, in terms of DGF, graft function, patient survival and graft survival, all at 1 year. Guarrera and colleagues’ calculations did not find these differences to be significant. However, our analysis indicated that the RR 1.05 (95%CI 1.01 to 1.08] was significant at p < 0.01 for patient and graft survival at 1 year. There were a large number of discarded kidneys following perfusion (25%); this may have been due to the high percentage of ECDs (78%).

Kazimi and colleagues’ much smaller study, of mostly better quality donor kidneys, found a non-significant gain in graft survival at 30 and 90 days for the RM3. They also found that people whose grafts had been stored in an RM3 had fewer days in hospital (RM3 = 3, LifePort = 15, p = 0.04). However, there were no differences in the number of times dialysis was needed post transplant. Post-storage/pre-transplant discard rates were similar (RM3 = 98, LifePort = 91).

Further robust research is needed using RCTs to determine the relative effectiveness of these perfusing machines.

Summary of cold storage solution versus cold storage solution

Three RCTs, one registry study and one hospital record review were found which compared the cold storage solutions of interest.

A multinational registry study compared ViaSpan with Marshall’s solution. Our analysis of the data showed that there were no significant differences between solutions for a range of CITs.

The three RCTs comparing ViaSpan with Celsior found no significant differences on any outcome measure; pooling these data continued to show no significant differences between groups.

The hospital record review, comparing ViaSpan with Celsior, only found a significant difference in creatinine concentrations at 1 and 12 months, with ViaSpan-stored kidneys having higher levels; these higher levels may have been due to the greater age of the recipients of those kidneys or other confounding factors not reported.

Post-storage/pre-transplant discard rates were similar (ViaSpan = 6, Celsior = 7).

Search strategy

The search strategy for economic evaluations and other economic studies is shown in Appendix 1. The range of sources searched is the same as for clinical effectiveness, with the addition of EconLit and NHS Economic Evaluation Database (EED).

Study selection criteria

The inclusion and exclusion criteria for the systematic review of economic evaluations were identical to those for the systematic review of clinical effectiveness, with the following exceptions:

• Decision model-based analyses or analyses of patient-level cost and effectiveness data alongside observational studies will be included.

• Full cost-effectiveness analyses, cost–utility analyses, cost–benefit analyses and cost–consequence analyses will be included. (Economic evaluations which report only average cost-effectiveness ratios will be included only if the incremental ratios can be easily calculated from the published data.)

• Stand-alone cost analyses based in the UK NHS will also be sought and appraised.

Based on the above inclusion/exclusion criteria, study selection was made by one reviewer (RA).

Study quality assessment

The methodological quality of the two included full economic evaluations has been assessed by an experienced health economist, partly by using the Consensus on Health Economics Criteria (CHEC) list questions developed by Evers and colleagues. 65

Summary of studies in our systematic review

Details of the key features and methods and the main results of the two included full economic evaluations are shown in Tables 17 and 18.

Wight and colleagues47 produced a systematic review of economic studies of machine perfusion of kidneys, and also reviewed research on the hypothetical relationship between DGF and graft survival. These reviews helped inform an original probabilistic cost–utility analysis of machine perfusion versus cold storage, which was directly based on a model of the relationship between DGF and graft survival using data from a single transplant centre in the US (from 1985 to 1990). 69

Their review of economic studies identified only three relevant studies (four articles), all of which were judged to be of poor quality. Two of the studies were not randomised and also reported that marginal kidneys were targeted to specified preservations systems. 70,71

In a more recent technology assessment report, for a Canadian university hospital research group, Costa and colleagues49 also examined the cost-effectiveness of machine perfusion versus cold storage of donated kidneys. Although in most respects this appears to be a relatively high quality model-based analysis, their cost-effectiveness results were expressed only in terms of the cost per DGF event avoided. As this can only be regarded as a surrogate outcome measure, the meaningfulness of their findings is somewhat limited. Furthermore, their analysis adopted a time horizon of only 1 year, and did not include any cost or other impacts of differential graft survival (and therefore any long-term changes in the pattern of life-years with a transplant as opposed to on dialysis).

Both studies predated the availability of effectiveness data from RCTs of machine perfusion versus cold storage.

Appendix 6 shows the extent to which each study satisfied different items in the CHEC criteria list for assessing the quality of economic evaluations. 65

Other relevant studies found

Two of the main purported benefits of better stored kidneys are that transplant recipients are less likely to need dialysis in the short term (i.e. lower rates of DGF), and may also need less dialysis in the longer term (i.e. because better stored kidneys may also have better long-term function and survival). Therefore, apart from the cost of machine perfusion or storage solutions themselves, the main economic (and quality of life) implication of better stored kidneys is reduced health care costs due to reduced patient life-years on dialysis. This happens to be the same main trade-off in economic evaluations which compare different forms of RRT or methods for expanding donor numbers. We examined several economic evaluations of alternative forms of RRT,38,72 methods for enhancing the kidney donor pool,73,74 alternative post-transplantation immunosuppressive regimes,75,76 or the economics of transplantation in general,64 in order to better understand the key trade-offs and how they might be estimated or simulated.

We also examined a number of economic studies which compared alternative methods of kidney storage. 67,68,70,77 Another older study, by Hornberger and colleagues,78 has also highlighted the potential importance for cost-effectiveness analyses of including re-transplantation as a treatment pathway.

Transitions between states

After each cycle of the model, patients are transferred from one state to another (or remain in the same state) according to the permitted transitions within the model. These transitions are represented by the arrows in the structure diagram of the model (see Figure 11). The probability of transferring from one state to another state is dependent on assigned transition probabilities which were derived from various sources and represent aspects of treatment effectiveness or natural disease progression (as described below). The full list of transitions represented in the model is shown in Appendix 7.

Pulsatile perfusion machines and solutions (LifePort only)

The cost of Waters’ RM3 machines to the NHS is not available (there was no industry submission for this machine and no transplant centres in the UK have bought this machine). A price was requested (via NICE) from the manufacturer, but was not supplied.

The purchase cost of a single LifePort machine is £10,750 (source: Organ Recovery Systems, budget impact analysis in submission to NICE, February 2008), but each transplant centre using machine perfusion would require two machines (one for each donated kidney), because kidneys are usually retrieved in pairs and each machine perfuses one kidney (total initial cost £21,500).

We have annualised this initial purchase cost, using the formula recommended by Drummond and colleagues. 82 In this calculation we have initially assumed that the LifePort technology (note, not each particular machine) will be used for 10 years in the NHS (before obsolescence or replacement by newer technologies). This is because, in addition to the initial purchase cost, most centres pay for a maintenance contract which replaces or repairs any broken or faulty machine (at an annual cost of US$1750 per machine). The annualised purchase cost therefore assumes a zero resale value after that time, the annuity factor for 10 years at 3.5% per year, and gives an annualised cost per LifePort machine of £1219, or £2438 for two machines. Transplant centres purchase two machines because usually two kidneys are retrieved from a deceased person. In addition, most UK centres currently using LifePort machines pay for an maintenance contract which costs US$1750 per machine (£874 using March 2008 exchange rates83) making the annual cost per machine £2092 (or £4184 for two machines). Finally, each LifePort-stored kidney also requires solutions and other consumables that are supplied as a perfusion kit (£475 each; source: Organ Recovery Systems, submission to NICE).

However, during any given year, the machines will be used for storing different numbers of kidneys in different transplant centres. Table 21 shows how the cost per kidney stored was calculated.

Cold storage boxes and solutions

In addition to the storage solutions, cold storage of kidneys involves the use of two sterile plastic bags, sterile ice, non-sterile ice and water, and non-sterile insulated boxes for storage and transportation. The boxes are bulk purchased and supplied to all transplant centres in the UK by the NHSBT. The vast majority are supplied with a satchel and the required accessories/consumables, costing £45.80 each (information supplied by the NHSBT); we use this figure in our base-case analyses. However, it should be noted that the current cost of replacement tubs with refill packs (i.e. without the satchel) is only £20.

Data supplied by the NHSBT indicate that 930 kidney boxes were supplied last year to transplant centres in the UK (figures for April 2007 to March 2008). Deducting an estimated 80 DCD kidneys which would have been stored using the LifePort machines (at eight transplant units) from the total of 1440 deceased donor transplants conducted in the UK in 2006–7, gives approximately 1360 kidneys which would have been stored using cold storage. This implies that each kidney storage box is used, on average, only 1½ times (i.e. 1360 ÷ 930), assuming all storage boxes are used up during this period.

Table 22 shows the cost per litre (excluding VAT) of the different storage solutions compared in our analysis.

Number and cost of kidney graft explantation

The NHSBT supplied data on the proportion of failed grafts which were explanted by time since transplant. Our assumptions regarding the probability of kidney graft explantation following graft failure are shown under Kidney graft failure, later in this chapter.

Each kidney explant operation is given a unit cost of £4135, which is the weighted average of the 2006–7 national average unit costs for kidney major open procedures (Healthcare Resource Group codes LB02B – with intermediate complex co-morbidities, and LB02C – without complex co-morbidities: £3949 and £4424 respectively).

Ongoing care costs with a functioning kidney transplant

Table 23 shows the main resource use assumptions and resultant monthly health-care costs we have included for those patients in the model with a functioning transplant.

Two transplant surgeons in our Expert Advisory Group suggested typical frequencies of outpatient appointments, which tend to reduce with time since transplant. The probability of acute rejection was also difficult to estimate because most studies only report short-term postoperative rates, which would overestimate long-term rates. We have therefore suggested simply reducing rates of acute reduction, with the initial rate for the first 3 months based on the rates reported in three of our included effectiveness studies.

For the cost of immunosuppression, in the absence of reliable national data on the exact drug protocols and doses used in all transplant centres, we relied on responses from our expert advisors (transplant surgeons) and NICE guidance. 84 We assumed that most transplant centres in the NHS use a triple regime involving (1) a calcineurin inhibitor (either ciclosporin or tacrolimus); (2) an antiproliferative agent (either azathioprine or mycophenolate mofetil; and (3) a steroid (usually prednisolone). We have not included the costs of initial ‘induction’ drug therapy (which is assumed to be incurred by all transplant recipients), and also have not specified lower immunosuppression costs for later years (as doses may be lowered over time).

With these ingredient costs, the estimated monthly NHS cost of living with a functioning transplant is initially £2464, decreasing to £1386, and then to £567 per month from year 2 onwards.

Ongoing care costs when on dialysis

Table 24 shows the main resource use assumptions and resultant monthly health-care costs we have included for those patients in the model who are on dialysis. As older patients are more likely to be on HD (rather than PD), we calculated age band-specific costs of being on dialysis to reflect how the costs of dialysis sessions and anaemia treatment would vary with age (Table 25).

Together these cost assumptions result in an average monthly cost of between £2034 and £2117, gradually increasing with patient age.

Costs not included

A more comprehensive analysis of the health-care cost of living with a transplant or on dialysis might include the following categories of resource use:

• GP visits/consultations and district nurse visits, which may differ between transplant patients and those on dialysis

• consultations with social care/social work professionals

• home adaptations (especially for people on HHD or on PD, e.g. showers, bunkers or sheds for storing deliveries of dialysate bags).

Summary of standard cost parameters

Table 26 lists the standard values of each of the cost variables used to calibrate the model.

Systematic search for comparative quality of life studies

Utility values used

Table 30 gives the utility values by age group for dialysis and transplant states in the model. The basis for these values is the age-related norms for the UK general population, to which a 0.1 decrement has been applied.

Immediate graft function/delayed graft function

The probabilities for immediate versus delayed graft function following transplant is a key parameter in the model and in general has been taken directly from the individual studies used in the model. The values used for each comparison are described at the beginning of each results section in this chapter.

Survival of functioning grafts

Graft survival was estimated using estimated graft survival curves which, in turn, were used to derive time-dependent probabilities for transition to the failing kidney states. In all cases, graft survival was modelled using Weibull curves, which were fitted to the trial data using regression analysis. For three of the four comparisons presented here, the study data presented gave a good initial basis for estimating the shape of the graft survival curves in each arm. However, in general, the study data did not provide sufficient length of follow-up to provide a high level of confidence around the fitted curves. In this context, therefore, we chose to use data provided by the NHSBT (Table 31) to extrapolate the curves to provide a more reliable fit. Also for one comparison, ViaSpan versus LifePort, the PPART trial did not provide graft survival data beyond 3 months post transplantation and showed no significant differences between arms. Therefore, in this case, we chose to use the the NHSBT graft survival data to fit the Weibull survival parameters for the model.

Kidney graft failure

Once graft failure occurs in the model, patients enter a failing kidney state where, within a very few cycles of the model (average 1.4 months), they are transferred to subsequent treatment by dialysis. The failing kidney model states have been introduced to reflect both the likely reduction in quality of life and higher associated treatment costs for patients whose kidney transplants are not functioning well, but who have not yet become dialysis dependent.

After graft failure, the model has two dialysis states: (1) receiving dialysis and waiting for further transplant; and (2) receiving dialysis unsuited to transplant. The relative probability of moving to each of the states is dependent on the age of the patient as outlined below.

Dialysis and re-transplantation following graft failure

Patients deemed suitable for re-transplantation following graft failure can receive subsequent (one or more) transplants in the model. This is represented using a single state which aggregates the costs, utilities and outcomes across all scenarios following re-transplant. The probability and waiting time for a patient receiving a subsequent transplant is known to be age related. Transition probabilities for re-transplant were therefore calculated independently for each age group based on data for the known numbers of re-transplant supplied by the NHSBT.

Patient survival

Renal Registry data81 were used to derive patient survival curves by age group and treatment modality (dialysis or transplant) for the standard data set used in the model. For those patients on dialysis, regression analysis was used to fit Weibull curves to Kaplan–Meier survival data for each of the age groups modelled (as shown in Figure 12).

FIGURE 12.

Standard patient survival curves by age group for patients on dialysis. [Source: UK Renal Registry 10th Annual Report (Figure 6.3b, p.100). The data reported here have been supplied by the UK Renal Registry of the Renal Association. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the UK Renal Registry or the Renal Association.]

Survival probability for patients on transplant is recognised to be significantly higher than for those on dialysis. An extensive analysis by Wolfe and colleagues2 revealed RR values of death across four differing age bands of patients ranging from 0.24 to 0.39. These data were confirmed by UK data supplied by the NHSBT for 5-year patient survival since transplant. To incorporate the improved survival of transplant patients relative to those on dialysis within the model, a HR of 0.327 was calculated as a weighted average based on the data presented by Wolfe and colleagues. This yielded the survival curves shown in Figure 13. Sensitivity analysis was used to explore the effects of changes to this HR on model outputs.

FIGURE 13.

Standard patient survival curves by age group for patients with transplant. (Source: UK Renal Registry 10th Annual Report. The data reported here have been supplied by the UK Renal Registry of the Renal Association. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the UK Renal Registry or the Renal Association.)

A summary of the parameters used in the PenTAG model is shown in Table 34.

LifePort versus ViaSpan

Two studies provide RCT data for the comparison of ViaSpan cold storage solution with LifePort machine perfusion. As these studies are based on different populations of both donor kidneys and recipients, and different trial conditions, each data set was modelled separately.

LifePort versus ViaSpan – PPART study with DCD kidney transplants in UK

In order to model cost–utility outcomes based on the PPART trial data, the standard data set was modified with the differential data shown in Table 35. For each of the arms, data were drawn from the reported trial outcomes and differential costs based on the resource analysis (described above).

In order to fit graft survival curves for these data in the model, it was necessary to use data supplied by the NHSBT for 5-year graft survival (classified by IGF and DGF) as the single 3-month data point provided by this trial does not provide a basis for survival curve fitting. The survival curves shown in Figure 14 were derived using the NHSBT data.

TABLE 35 - Summary of differential input parameters based on PPART trial data

DGF, delayed graft function; PNF, primary non-function.

Parameter	ViaSpan	LifePort
Storage cost per kidney	£262.53	£736.55
Percentage of DGF following transplant	(Academic-in-confidence information removed)	(Academic-in-confidence information removed)
Percentage of PNF	(Academic-in-confidence information removed)	(Academic-in-confidence information removed)
Graft survival (all patients) at 3 months	(Academic-in-confidence information removed)	(Academic-in-confidence information removed)

FIGURE 14.

Weibull survival estimates of graft survival for IGF and DGF patient groups used by the model for comparison of ViaSpan and LifePort based on the PPART trial.

These data yielded the summary deterministic outputs from the model for cost and benefit differences shown in Table 36.

TABLE 36 - Base-case deterministic outputs from PenTAG model based on PPART trial data

ICER, incremental cost-effectiveness ratio; QALYs, quality-adjusted life-years.

	Discounted costs per patient (£)	Discounted benefits per patient (QALYs)	ICER
ViaSpan cold storage	139,205	9.19
LifePort machine perfusion	141,319	9.13	Was dominated
Difference	2114	–0.066
	Undiscounted costs per patient (£)	Undiscounted benefits per patient (QALYs)	ICER
ViaSpan cold storage	228,885	16.51
LifePort machine perfusion	231,387	16.36	Was dominated
Difference	2502	–0.153

The outputs from the model show only very small differences between the arms for both costs and benefits. This reflects the fact that there are only very small differences in the rates of DGF and PNF. However, LifePort was dominated by ViaSpan, i.e. ViaSpan both was less costly and produced more benefits than LifePort. Appendix 8 shows the breakdown of these results by age group.

N.B. When uncertainty about the effectiveness estimates is factored into these inputs it is difficult to arrive at any firm conclusion about a preferred storage alternative based on these trial data.

The component analyses in Figures 15 and 16 show how the incremental costs and benefits between the comparator arms are broken down in terms of their contributory elements. They show that the cost increases from the overall higher lifetime dialysis requirements are higher than any savings associated with reduced survival (LifePort confers slightly less patient survival so there is an associated cost saving).

FIGURE 15.

Component analysis of incremental cost of LifePort vs ViaSpan (PPART data).

FIGURE 16.

Component analysis of utility gains LifePort vs ViaSpan. QALY(s), quality-adjusted life-year(s).

The component analysis in Figure 16 shows that most of the estimated reduction in QALYs with LifePort were due to reduced patient survival (in turn due to more life-years on dialysis), and only partly due to the reduced quality of life when on dialysis.

The event counts that were output by the model for this comparison for a cohort of 1000 simulated kidney graft recipients are shown in Table 37.

TABLE 37 - Events count output from PenTAG model based on PPART trial data

Description	ViaSpan	LifePort
Immediate graft function (IGF)	444	422
Delayed graft function (DGF)	556	578
Primary non-function	0	22
Graft failures after IGF	60	57
Deaths in IGF	366	347
Graft failures after DGF	181	181
Deaths in DGF	362	362
Explants after graft failure	18	18
Non-explant after graft failure	219	216
Waiting list after graft failure	96	95
Unsuitable for transplant after graft failure	141	139
Re-transplants	97	119
Graft failures after re-transplant	66	81
Deaths in subsequent transplant	29	36
Deaths while waiting for re-transplant	64	77
Deaths on dialysis (transplant unsuited)	140	138

LifePort versus ViaSpan – the MPT in BSD and DCD patients in Germany, Belgium and the Netherlands

In order to model cost–utility outcomes based on the MPT data, the standard data set was modified with the data drawn from the costing assumptions (described above) and the reported trial outcomes (Table 38).

TABLE 38 - Summary of differential input parameters based on MPT data

DGF, delayed graft function; IGF, immediate graft function; PNF, primary non-function.

Parameter	ViaSpan	LifePort
Storage cost per kidney	£262.53	£736.55
Proportion of DGF following transplant	26.5%	20.8%
Proportion of PNF	4.8%	2.1%
Graft survival (IGF patients)	98% at 1 year	98% at 1 year
Graft survival (DGF patients)	82% at 1 year	93% at 1 year

For the graft survival in the model, regression analysis was used to fit a Weibull curve for the graft survival parameters. In order to provide a representative fit, data supplied by the NHSBT for 5-year graft survival (classified by IGF and DGF) were used to extrapolate the HR for each population beyond the first year supplied in the trial data. This yielded the survival curves shown in Figure 17. It should be noted here that it was necessary to read survival estimates directly from presented Kaplan–Meier curves, permitting possible error. It would have been useful to have the corresponding numerical data for graft survival from this trial in accordance with best practice for presenting survival data. 116

FIGURE 17.

Weibull survival estimates of graft survival for IGF and DGF patient groups used by the model for comparison of ViaSpan and LifePort based on the MPT.

Table 39 shows the base-case outputs from the model for each comparator arm. These are the deterministic model outputs with discounting and show the cost and utilities per patient for each treatment option, as well as the incremental values for costs and QALYs.

TABLE 39 - Base-case deterministic outputs from PenTAG model based on MPT data

ICER; incremental cost-effectiveness ratio; QALYs, quality-adjusted life-years.

	Discounted costs per patient (£)	Discounted benefits per patient (QALYs)	ICER
ViaSpan cold storage	142,805	9.58	Was dominated
LifePort machine perfusion	139,110	9.79
Difference	–3695	0.218
	Undiscounted costs per patient (£)	Undiscounted benefits per patient (QALYs)	ICER
ViaSpan cold storage	232,301	17.20	Was dominated
LifePort machine perfusion	228,540	17.68
Difference	–3761	0.485

The deterministic outputs from the model show that, for the input parameters derived from this study, LifePort machine perfusion dominates the cost–utility analysis. That is to say that this method of storage results in both lower overall costs of treatment and greater benefits to patients when compared with cold storage using the ViaSpan solution. Appendix 8 shows the breakdown of these results by age group.

The component analyses in Figures 18 and 19 show how the difference in costs and benefits between the comparator arms is broken down. Here it can be seen that the cost savings from reducing the dialysis requirement far outweighs both the costs associated with kidney storage and those associated with increased survival (N.B. increased survival is associated with extra costs because either being on dialysis or having a working kidney graft incurs significant ongoing health-care costs).

FIGURE 18.

Component analysis of incremental costs of LifePort vs ViaSpan (MPT data).

FIGURE 19.

Component analysis of utility gains of LifePort vs ViaSpan. QALY(s) quality-adjusted life-year(s).

In Figure 19 it can be seen that, compared with ViaSpan, LifePort machine preservation confers additional QALYs, mainly through survival gains rather than through the utility gains associated with less time back on dialysis.

The event counts in Table 40 were output by the model for this comparison for a cohort of 1000 simulated kidney graft recipients.

TABLE 40 - Events count output from PenTAG model based on MPT data

Description	ViaSpan	LifePort
Immediate graft function (IGF)	735	792
Delayed graft function (DGF)	265	208
Primary non-function	48	21
Graft failures after IGF	93	100
Deaths in IGF	612	660
Graft failures after DGF	132	103
Deaths in DGF	84	83
Explants after graft failure	18	14
Non-explant after graft failure	203	186
Waiting list after graft failure	89	81
Unsuitable for transplant after graft failure	132	119
Re-transplants	142	103
Graft failures after re-transplant	97	70
Deaths in subsequent transplant	44	31
Deaths while waiting for re-transplant	90	68
Deaths on dialysis (transplant unsuited)	131	118

LifePort versus Marshall’s Soltran

For the cost–utility comparison of Marshall’s Soltran solution versus LifePort MP, one clinical effectiveness study by Plata-Munoz and colleagues55 has been used to provide effectiveness data for this cost–utility analysis. The comparator-specific data shown in Table 41 were input into the model in addition to the standard data set described above.

Regression analysis was used to fit a Weibull curve for each of the graft survival parameters used for this comparison. In order to provide a representative fit, data supplied by the NHSBT for 5-year graft survival (classified by IGF and DGF) were used to extrapolate beyond the 2-year results supplied in the trial data. No data were supplied in the trial to discriminate between graft survival for IGF and DGF patients, so both population groups were assumed to experience the same graft survival. Figure 20 shows the survival curves for each arm that were employed in the model.

FIGURE 20.

Weibull survival estimates of graft survival for each arm of comparison of Marshall’s Soltran and LifePort.

These data yielded the summary base-case outputs from the model for cost and benefit differences shown in Table 42.

The deterministic outputs from the model show that, for the input parameters derived from this study, LifePort machine perfusion dominates the cost–utility analysis. That is to say that this method of storage results in both lower overall costs of treatment and greater benefits to patients when compared with cold storage using the Marshall’s solution. Appendix 8 shows the breakdown of these results by age group.

The component analyses in Figures 21 and 22 show the breakdown of costs and utility gains between the comparator arms. These figures show that the reduction in dialysis costs is the most important factor in the relatively lower costs of LifePort, and that improved graft survival is the key factor leading to the greater QALY output for LifePort compared with Marshall’s Soltran.

FIGURE 21.

Component analysis of incremental costs of LifePort vs Marshall’s Soltran.

FIGURE 22.

Component analysis of incremental utility gains of LifePort vs Marshall’s Soltran. QALY(s), quality-adjusted life-year(s).

The event counts in Table 43 were output by the model for this comparison for a cohort of 1000 simulated kidney graft recipients.

ViaSpan versus Marshall’s Soltran solution

For the cost–utility comparison of Marshall’s Soltran solution versus LifePort, one study (Opelz and Dohler57) satisfied our inclusion criteria. This registry data study provided inputs for graft survival at 3 years. The between-arms data shown in Table 44 were put into the model in addition to the underlying standard data set.

Weibull curve fits for each of the graft survival parameters used for this comparison were calculated using regression analysis. Three-year graft survival data for each arm were extracted from the study data and used to calculate representative Weibull parameters for each arm of the trial. As no data were supplied to distinguish between graft survival for IGF versus DGF patients in this study, both patient groups were assumed to have the same graft survival. The survival curves for each arm shown in Figure 23 were employed in the model. For many data points, it was necessary to read survival estimates directly from presented Kaplan–Meier curves and it would have been useful to have the corresponding numerical data for graft survival from this trial in accordance with best practice for presenting survival data. 116

FIGURE 23.

Weibull survival estimates of graft survival for each arm of comparison of ViaSpan and Marshall’s Soltran.

Table 45 shows the summary base-case outputs from the model; these indicate that for the specific input parameters derived from this study, ViaSpan both results in lower overall costs of treatment and confers greater benefits to patients when compared with cold storage using the Marshall’s Soltran solution. However, these differences are seen to be very small in the context of the overall levels of uncertainty surrounding the input parameters. In practice, it is difficult to make conclusions based on these output data with any level of confidence. Appendix 8 shows the breakdown of these results by age group.

The following component analyses in Figures 24 and 25 show the breakdown of costs and benefits between the comparator arms. This again shows that it is the costs of dialysis that are having the major influence on cost outcomes, together with gains in survival from ViaSpan, causing it to dominate Marshall’s Soltran.

FIGURE 24.

Component analysis of incremental costs of Marshall’s Soltran vs ViaSpan.

FIGURE 25.

Component analysis of incremental benefits of Marshall’s Soltran vs ViaSpan. QALY(s), quality-adjusted life-year(s).

The event counts in Table 46 were output by the model for this comparison for a cohort of 1000 simulated kidney graft recipients.

LifePort versus ViaSpan

LifePort versus ViaSpan – PPART study with DCD donor kidney transplants

Figure 31 shows the scatter plot outputs from the model for 1000 trial runs of the probabilistic simulation. These demonstrate the levels of uncertainty associated with the cost and effectiveness outputs from both arms of this comparison, when the parameter uncertainty is included in the model. The figure shows that the variation due to parameter uncertainty within each arm is much greater than any difference between the arms.

FIGURE 31.

Scatter plot from probabilistic simulation based on PPART data for ViaSpan vs LifePort. QALYs, quality-adjusted life-years.

Figure 32 represents the outputs shown in Figure 31 in terms of the incremental costs and benefits of LifePort versus ViaSpan. Net benefit thresholds are shown for willingness-to-pay thresholds of £20,000 per QALY (solid line) and £30,000 per QALY (dashed line). Once again, the inherent uncertainty of the outputs is shown by the distribution of dots across the cost-effectiveness plane. This graph shows that there is no clear conclusion that can be drawn about the relative cost-effectiveness.

Figure 33 shows the CEAC for the comparison of ViaSpan with LifePort based on the PPART data. This shows the probability, based on the probabilistic model outputs, that the LifePort storage option is cost-effective over a range of different levels of willingness to pay for each extra QALY conferred by adopting this treatment. This, in turn, shows that over a range of willingness-to-pay thresholds the model predicts around a 40% likelihood that LifePort will be cost-effective when compared with ViaSpan.

FIGURE 32.

Incremental cost-effectiveness ratio of LifePort vs ViaSpan based on PPART trial data. QALYs, quality-adjusted life-years.

FIGURE 33.

Cost-effectiveness acceptability curve for LifePort vs ViaSpan: PPART trial data. QALY, quality-adjusted life-year.

LifePort versus ViaSpan – MPT in BSD and DCD patients

Figure 34 shows the scatter plot outputs from the model for 1000 trial runs of the probabilistic simulation based on the inputs from the MPT data. Levels of uncertainty associated with the cost and effectiveness outputs from both arms of this comparison, when the parameter uncertainty is included in the model, are demonstrated by the distribution of output points. The scatter plot shows that the estimated cost-effectiveness of the comparators is very similar.

FIGURE 34.

Scatter plot from probabilistic simulation based on MPT data for LifePort vs ViaSpan. QALYs, quality-adjusted life-years.

Figure 35 represents the outputs shown above in terms of the incremental costs and benefits of LifePort versus ViaSpan. Net benefit thresholds are shown for willingness-to-pay thresholds of £20,000 per QALY (solid line) and £30,000 per QALY (dashed line). Once again, the inherent uncertainty of the outputs is shown by the distribution of dots across the cost-effectiveness plane. The majority of data points in the lower right-hand quadrant indicates that LifePort is more likely to be cost-effective at any level of willingness to pay.

FIGURE 35.

Incremental cost-effectiveness ratio of LifePort MP vs ViaSpan based on MPT data. QALYs, quality-adjusted life-years.

Figure 36 shows the CEAC for the comparison of LifePort with ViaSpan based on the MPT data. This shows the probability based on the PSA outputs that the LifePort storage option is cost-effective over a range of different levels of willingness to pay for each extra QALY conferred by adopting this treatment. It indicates that there is an 80% probability that LifePort is cost-effective across the willingness-to-pay range.

FIGURE 36.

Cost-effectiveness acceptability curve for LifePort vs ViaSpan based on MPT data. QALY, quality-adjusted life-year.

LifePort versus Marshall’s Soltran

Figure 37 shows the scatter plot outputs from the model for 1000 trial runs of the probabilistic simulation based on the trial data for cold storage with Marshall’s solution versus LifePort machine preservation. The distribution of output points illustrates the levels of uncertainty associated with the cost and effectiveness outputs from both arms of this comparison, when the parameter uncertainty is taken into account in the model. This illustrates the large level of uncertainty apparent in model outputs when parameter uncertainty is incorporated. Once again, there is a strong overlap between the outputs from each arm, indicating much more variation within the comparator arms than between them.

FIGURE 37.

Scatter plot from probabilistic simulation for Marshall’s Soltran vs LifePort. CS, cold storage; MP, machine perfusion; QALYs, quality-adjusted life-years.

Figure 38 represents the outputs shown above in terms of the incremental costs and benefits of LifePort versus Marshall’s Soltran. Net benefit thresholds are shown for willingness-to-pay thresholds of £20,000 per QALY (solid line) and £30,000 per QALY (dashed line). This shows that for these data there is a high level of uncertainty inherent in the output simulations, with LifePort dominating over Marshall’s Soltran in a great number of the simulation trials.

FIGURE 38.

Incremental cost-effectiveness of LifePort vs Marshall’s Soltran. QALYs, quality-adjusted life-years.

Figure 39 shows the CEAC for the comparison of Marshall’s Soltran with LifePort. This shows that LifePort is estimated to have a greater than 95% probability of being more cost-effective than Marshall’s Soltran for this data set for a large range of willingness-to-pay thresholds. However, these are not RCT data and these outputs should be treated with caution.

FIGURE 39.

Cost-effectiveness acceptability curve for LifePort vs Marshall’s Soltran. QALYs, quality-adjusted life-years.

ViaSpan versus Marshall’s Soltran

Figure 40 shows the scatter plot outputs from the model for 1000 trial runs of the probabilistic simulation based on the trial data for cold storage with Marshall’s solution versus LifePort machine preservation. The distribution of output points illustrates the levels of uncertainty associated with the cost and effectiveness outputs from both arms of this comparison, when the parameter uncertainty is taken into account in the model. Once again, this distribution shows that the within-comparator variation is much greater than the between-comparator variation, once parameter uncertainty is incorporated into the model.

FIGURE 40.

Scatter plot from probabilistic simulation for Marshall’s Soltran vs ViaSpan. QALYs, quality-adjusted life-years.

Figure 41 represents the outputs in Figure 40 in terms of the incremental costs and benefits of LifePort versus Marshall’s Soltran. Net benefit thresholds are shown for willingness-to-pay thresholds of £20,000 per QALY (solid line) and £30,000 per QALY (dashed line). This graph shows that, based on the data from this study, there is very little to distinguish between the cost-effectiveness of Marshall’s Soltran and that of ViaSpan. It should be noted that these outputs are based on a single study.

FIGURE 41.

Incremental cost-effectiveness of Marshall’s Soltran vs ViaSpan. QALYs, quality-adjusted life-years.

Figure 42 shows the CEAC for the comparison of Marshall’s Soltran with ViaSpan. This graph shows around a 40% probability that Marshall’s Soltran is cost-effective when compared with ViaSpan across a wide range of willingness-to-pay thresholds. Hence, there is little in these outputs to help us to determine cost-effectiveness between the two comparators.

FIGURE 42.

Cost-effectiveness acceptability curve for Marshall’s Soltran vs ViaSpan. QALY, quality-adjusted life-year.

Machine perfusion versus cold storage

Unfortunately we are unable to provide a clear answer to the issue of comparing machine perfusion with cold storage in DCD kidneys. There were two recent RCTs of this comparison, one of which (PPART, n = 90) (academic-in-confidence information removed), while the other (MPT, n = 672) produced a non-significant result in favour of machine perfusion for most short-term outcomes (e.g. DGF, PNF), but showed a small significant difference in graft survival at 12 months in favour of machine perfusion (HR at 12 months 0.39, p = 0.03). However, there are a number of important differences between these two trials, in terms of the kidney donor types, study design and settings, and the integrity of the actual interventions received, which may explain some of the differences in their results.

The PPART RCT solely used DCD donor kidneys (at five transplant centres in the UK) (academic-in-confidence information removed).

In contrast, the MPT included mostly BSD and some DCD donor kidneys (294 BSD, 42 DCD). Although there was no significant difference in DGF or PNF between the two storage methods, their 12-month follow-up results indicate improved graft survival with machine perfusion. A subgroup analysis included additionally-recruited DCD participants (n = 82), but also failed to show any difference in DGF (their designated primary outcome) between types of donor. It would therefore be speculative to extrapolate the full trial findings, using largely BSD donor kidneys, to DCD kidneys and their recipients. From the MPT, for the key outcome of 1-year graft survival it does seem that there may be an advantage from machine perfusion in BSD-donated kidneys. Other outcomes fail to show a benefit for either storage method, at least when CIT is between about 3 and 30 hours (mean 15 hours). It is paradoxical that this trial, with mostly BSD kidneys, gives stronger indications of the relative effectiveness of machine perfusion than the trial in DCD kidneys (which would, theoretically, be expected to benefit more from the technology). This result may not hold with longer cold ischaemic or at post-transplant follow-up times.

The only study we found comparing LifePort with Marshall’s Soltran (Plata-Munoz and colleagues55) had many potentially confounding factors: it wasn’t randomised; for the first 2 years all kidneys were perfused with Marshall’s Soltran and subsequently machine preservation with LifePort was used; the size of the study was small (n = 60); the mean age of recipients of kidneys that had been cold stored was 7 years older that those stored with LifePort; and kidneys stored with LifePort had a longer CIT. Taken together, these factors mean that very little credence can be given to this study’s results.

Effectiveness of different kidney perfusion machines

The lack of any RCT or fully published evidence makes it very difficult to say whether either of the two machines assessed is better. However, the two record review studies that we found suggest that the RM3 may perform better than LifePort. These results may have been subject to confounding influences; well-designed RCTs are needed to establish if either machine is better.

Effectiveness of different cold storage solutions

The results from the RCTs comparing cold storage solutions indicate that, at least for CIT of less than approximately 15 hours, ViaSpan and Celsior are equivalent for kidney preservation. Registry evidence suggests that there is no significant difference between ViaSpan and Marshall’s Soltran for graft survival for a range of CITs.

The conclusions that this systematic review can come to are uncertain and limited by the lack of RCTs, and the number of studies that have not finished collecting and analysing their results and/or have not published them fully.

No existing systematic reviews that met our assessment criteria were identified.

Summary of previously published economic evaluations

There were only two previously published economic evaluations which met the inclusion criteria of our systematic review. The analysis by Wight and colleagues,47 while fairly recent and conducted from a UK NHS perspective, was not able to make use of the two most recent RCTs of machine perfusion versus cold storage of donated kidneys. Also, its results were highly dependent on an estimated relationship between DGF and graft survival, which we think is no longer defensible (given both mixed evidence about the existence of this relationship, and recent trials reporting graft and patient survival as pre-specified outcomes). The other economic evaluation, by Costa and colleagues,49 was conducted from a Canadian university hospital perspective, and had a number of important shortcomings in relation to the quality of the study and its relevance to the present decision problem.

Summary of PenTAG’s model-based cost–utility analysis

We were able to model the lifetime cost and QALY impacts of: machine perfusion with LifePort versus cold storage with ViaSpan; machine perfusion with LifePort versus cold storage with Marshall’s Soltran; and cold storage with ViaSpan versus cold storage with Marshall’s Soltran. In each case, however, the base-case deterministic results should be viewed with considerable caution, owing to both the uncertainty surrounding the relevant clinical effectiveness study results, and also the uncertainty surrounding whether short-term differences in graft survival (between different storage methods) would be manifested in the longer term.

Timing of assessment

• We have not had the 12-month follow-up data from the PPART trial, which, although weakened by the conduct of the study (see above), is the only RCT that has compared hypothermic machine perfusion with cold storage in DCD donors. Although the MPT included DCD donors (n = 84), this trial was predominantly of BSD donors (n = 588). Subgroup analysis only examined DGF.

• Additionally, the only studies found that compared the two preservation machines have not yet been published as peer-reviewed articles. This has the effect of limiting the information that can be gleaned about the conduct and outcomes of this research.

• The effects of this limitation are that we cannot be sure of the long-term effects on graft and patient survival of mode of kidney storage, especially as no significant differences were found in DGF or PNF. We also have little insight into the relative merits of the two preservation machines.

Quality of effectiveness studies

• Only 5 of the 11 included studies were RCTs; this meant that some of the comparisons (LifePort versus RM3, LifePort versus Marshall’s Soltran and Marshall’s versus ViaSpan) were dependent on data from studies where, due to less robust design, there may have been selection and other biases, possibly confounding the results.

• (Academic-in-confidence information removed.) We do not know what effect this may have had on the results.

Scope

• As the manufacturers of Celsior (Genzyme) were not invited to make a submission for this assessment, it has not been possible to include Celsior in the cost-effectiveness analysis. This is a shame as the pooled results of the three RCTs comparing Celsior with ViaSpan indicate their equivalence.

Competing interests of Expert Advisory Group

Mr Chris Watson is the Principal Investigator of the PPART trial. No other competing interests were declared in relation to this assessment.

Cochrane Library (CDSR and CENTRAL)

Wiley: online version 2007, issue 4

Search date: 29 November 2007

1. #1. MeSH descriptor Kidney Transplantation, this term only

2. #2. MeSH descriptor Tissue Donors, this term only

3. #3. MeSH descriptor Organ Preservation Solutions, this term only

4. #4. MeSH descriptor Organ Preservation, this term only

5. #5. MeSH descriptor Tissue Preservation, this term only

6. #6. kidney* OR renal*

7. #7. MeSH descriptor Kidney explode all trees

8. #8. (#6 OR #7)

9. #9. (#2 OR #3 OR #4 OR #5)

10. #10. (#8 AND #9)

11. #11. (#1 OR #10)

12. #12. MeSH descriptor Pulsatile Flow, this term only

13. #13. MeSH descriptor Perfusion, this term only

14. #14. (machine or pulsat*)

15. #15. (#13 AND #14)

16. #16. lifeport

17. #17. (machine or pulsat*) NEAR (Perfusion)

18. #18. RM3

19. #19. (machine or pulsat*) NEAR (perfus* or preserv* or system)

20. #20. ((cold or ice or static) AND (storag* or preserv*)):ti,ab

21. #21. eurocollins

22. #22. HTK

23. #23. histidine and tryptophan

24. #24. celsior

25. #25. viaspan

26. #26. soltran

27. #27. (university NEAR wisconsin):ti,ab

28. #28. belzer*

29. #29. (#12 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #27 OR #28)

30. #30. (#29 AND #11)

MEDLINE (1950 to date)

Dialog DataStar: online version

Search date: 29 November 2007

1. KIDNEY-TRANSPLANTATION#.DE.

2. (RENAL OR KIDNEY$3) NEAR (TRANSPLANT$6 OR PRESERV$OR REPLACE$OR DONOR$OR DONOUR$OR DONATE$OR RECIEVE$)

3. TISSUE-DONORS#.DE. OR ORGAN-PRESERVATION-SOLUTIONS.DE. OR ORGAN-PRESERVATION.DE. OR TISSUE-PRESERVATION#.DE.

4. KIDNEY.W..MJ.

5. KIDNEY$3 OR RENAL

6. 4 OR 5

7. ADJ ORGAN ADJ TRANSPLANT$6).TW.

8. (NO

9. 6 AND 3

10. 1 OR 2 OR 7

11. (SOLID N-HEART-BEATING OR NON ADJ HEART ADJ BEATING OR NHBD OR HEART ADJ BEATING OR HEART-BEATING OR CADAV$4 OR BRAIN ADJ DEAD).TW.

12. (DONOR$2 OR DONOUR$2) NEAR (MARGINAL OR EXPANDED OR EXTENDED OR HIGH-RISK)

13. 9 OR 10 OR 11

14. 12 AND 6

15. 13 OR 8

16. PULSATILE-FLOW#.DE.

17. MACHINE$2.TW. AND PULSAT$4.TW.

18. LIFEPORT.TW.

19. RM3.TI,AB.

20. (MACHINE$2 OR PULSAT$4).TW. AND (PERFUS$4 OR PRESERV$4 OR SYSTEM).TW.

21. WATER$2 ADJ RM3

22. KIDNEY.W..MJ.OR RENAL OR KIDNEY$3

23. WATER$2 NEAR PRESERVATION AND 21

24. WATER$2 ADJ MEDICAL ADJ SYSTEM$2

25. WATER$2 NEXT RENAL$2

26. 24 AND 21

27. KIDNEY$2 NEXT TRANSPORT$4 AND 22

28. UNIVERSITY ADJ OF ADJ WISCONSIN OR UW ADJ SOLUTION$2

29. CELSIOR

30. MARSHALL’S NEAR SOLUTION

31. VIASPAN

32. SOLTRAN

33. BELZER$

34. PERFUSION#.W..DE. AND (machine OR pulsat$4).TW.

35. (cold OR ice OR static OR hypo OR thermic).TI,AB. AND (storage OR preserv$5).TI,AB

36. (histidine AND tryptophan) OR HTK

37. 15 OR 16 OR 17 OR 18 OR 19 OR 20 OR 21 OR 22 OR 23 OR 24 OR 26 OR 27 OR 28 OR 29 OR 30 OR 31 OR 32 OR 33 OR 34 OR 35

38. 36 AND 14

39. LG=EN

40. 37 AND 38

41. PT=EDITORIAL OR PT=LETTER

42. ANIMAL=YES NOT HUMAN ADJ =YES

43. NOT (40 OR 41)

EMBASE (1974 to date)

Dialog DataStar: online version

Search date: 29 November 2007

1. KIDNEY-TRANSPLANTATION#.DE.

2. ((KIDNEY$3 OR RENAL) NEAR (TRANSPLANT$6 OR PRESERV$5 OR REPLACE$6 OR DONOR$2 OR DONOURS$2 OR DONAT$3 OR RECEIVE$4)).TI,AB.

3. ORGAN-DONOR.MJ.

4. KIDNEY-DONOR.MJ.

5. KIDNEY-PRESERVATION.MJ.

6. ORGAN-PRESERVATION.MJ.

7. PRESERVATION-SOLUTION#.DE.

8. TISSUE-PRESERVATION#.DE.

9. ((DONOR$2 OR DONOUR$2) NEAR (MARGINAL OR EXPANDED OR EXTENDED OR HIGH-RISK)).TI,AB.

10. (NON-HEART-BEATING OR NON ADJ HEART ADJ BEATING OR HEART-BEATING OR HEART ADJ BEATING).TI,AB.

11. (SOLID ADJ ORGAN ADJ TRANSPLANT$6).TI,AB.

12. KIDNEY#.W..DE.

13. KIDNEY$3 OR RENAL

14. 12 OR 13

15. 1 OR 2 OR 4 OR 5

16. 3 OR 6 OR 7 OR 8 OR 9 OR 10 OR 11

17. 16 AND 14

18. 15 OR 17

19. PULSATILE-FLOW#.DE.

20. KIDNEY-PERFUSION.MJ.

21. PERFUSION#.W..DE.

22. 21 AND (MACHINE OR PULSAT$4)

23. LIFEPORT.TW.

24. RM3.TI,AB.

25. (12 OR 13) AND (MACHINE$2 OR PULSAT$4) AND (PERFUS$4 OR PRESERV$4 OR SYSTEM)

26. (12 OR 13) AND (UNIVERSITY ADJ OF ADJ WISCONSIN OR UW ADJ SOLUTION)

27. CELSIOR

28. MARSHALL’S NEAR SOLUTION

29. VIASPAN

30. SOLTRAN

31. BELZER$

32. HISTIDINE AND TRYPTOPHAN OR HTK

33. (COLD OR ICE OR STATIC OR HYPO OR THERMIC).TI,AB. AND (STORAGE OR PRESERV$5).TI,AB.

34. MACHINE$2 AND PULSAT$4

35. 19 OR 20 OR 22 OR 23 OR 24 OR 25 OR 26 OR 27 OR 28 OR 29 OR 30 OR 31 OR 32 OR 33 OR 34

36. 35 AND 18

37. LG=EN

38. AT=EDITORIAL OR AT=LETTER

39. ANIMAL=YES NOT HUMAN=YES

40. 38 OR 39

41. 36 AND 37

42. 41 NOT 40

CINAHL (1982 to date)

Dialog DataStar: online version

Search date: 29 November 2007

1. (RENAL OR KIDNEY$3) NEAR (TRANSPLANT$6 OR PRESERV$6 OR REPLACE$OR DONOR$5 OR DONOUR$5 OR DONATE$OR RECEIVE$).TI,AB.

2. KIDNEY-TRANSPLANTATION#.DE.

3. ORGAN-PRESERVATION#.DE.

4. TRANSPLANT-DONORS#.DE.

5. (SOLID ADJ ORGAN NEAR TRANSPLANT).TI,AB.

6. (NON-HEART-BEATING OR NON-HEART OR HEART-BEATING OR NHBD OR HEART ADJ BEATING OR CADAV$4 OR BRAIN ADJ DEAD).TI,AB.

7. (DONOR$4 OR DONOUR$4) NEAR (MARGINAL OR EXPANDED OR EXTENDED OR HIGH-RISK)

8. KIDNEY#.W..DE.

9. (KIDNEY$3 OR RENAL).TI,AB.

10. 8 OR 9

11. 3 OR4 OR 5 OR 6 OR 7

12. 10 AND 11

13. 12 OR 1 OR 2

14. (MACHINE$2 OR PULSAT$4).TI,AB. AND (PERFUS$4 OR PRESER$4 OR SYSTEM).TI,AB.

15. UNIVERSITY ADJ OF ADJ WISCONSIN OR UW ADJ SOLUTION$

16. LIFEPORT OR RM3

17. CELSIOR OR VIASPAN OR SOLTRAN OR BELZER$

18. MARSHALL$NEAR SOLUTION$

19. MACHINE AND PULSATILE

20. (10 OR 2) AND KIDNEY$3 NEXT TRANSPORT$4

21. WATER$2 NEXT RENAL$2 OR WATER$2 NEAR PRESERVATION

22. (21 OR 19) AND (2 OR 10)

23. 13 AND (14 OR 15 OR 16 OR 17 OR 18 OR 20 OR 22)

24. 23 AND LG=EN

25. PT=BIBLIOGRAPHY OR PT=CEU OR PT=COMMENTARY OR PT=EDITORIAL OR PT=EXAM-QUESTIONS OR PT=GLOSSARY OR PT=LETTER OR PT=OBITUARY

26. 24 NOT 25

ISI Web of Knowledge (SCI-Expanded) (1970 to date)

Search date: 28 November 2007

1. #1. TS=((university SAME wisconsin) OR (UW SAME solution)))

2. #2. TS=((histidine SAME tryptophan) OR (marshall* SAME solution))

3. #3. TS=(HTK or celsior or viaspan or soltran or belzer*)

4. #4. TS=((machine or pulsat* or perfus*) AND (preserv* or system or storage*))

5. #5. TS=((machine) AND (pulsat* or perfus*))

6. #6. TS=((cold or ice or static or therm*) AND (storage or preserv*))

7. #7. #6 OR #5 OR #4 OR #3 OR #2 OR #1

8. #8. TS=((kidney* or renal*) AND (preserv* or replace* or donor* or donour* or receive* or transplant* or procurement))

9. #9. #8 AND #7

10. #10. #9 AND Language=(English)

11. #11. TI=(rat* or porcin* or canin*) AND Language=(English)

12. #12. #10 not #11 AND Language=(English)

ISI Web of Knowledge (ISI Proceedings, Science & Technology edition) (1990 to date)

Years searched: 2003 to date

Search date: 27 November 2007

1. #1. TS=((university same wisconsin) OR (UW same solution) or (histidine SAME tryptophan) OR (marshall* SAME solution))

2. #2. TS=((eurocollins or HTK or celsior or viaspan or soltran or belzer*))

3. #3. TS=((machine or pulsat* or perfus*) AND (preserv* or system or storage*))

4. #4. TS=((machine) AND (pulsat* or preserv*))

5. #5. TS=((kidney* or renal*) AND (preserv* or replace* or donour* or donor* or receive* or transplant* or procurement))

6. #6. #4 OR #3 OR #2 OR #1

7. #7. #6 AND #5

8. #8. #7 AND Language=(English)

9. #9. TI=(rat* or porcin* or canin*)

10. #10. #8 not #9

11. #11. #10 (Databases=STP Timespan=2003–2007)

Database of Abstracts of Reviews of Effects (DARE) on the CRD website

Search date: 29 November 2007

1. #1. MeSH Kidney Transplantation

2. #2. MeSH Tissue Donors

3. #3. MeSH Organ Preservation Solutions

4. #4. MeSH Organ Preservation

5. #5. MeSH Tissue Preservation EXPLODE 3

6. #6. kidney* OR renal

7. #7. MeSH Kidney

8. #8. #6 OR #7

9. #9. #2 OR #3 OR #4 OR #5

10. #10. #8 AND #9

11. #11. MeSH Pulsatile Flow

12. #12. MeSH Perfusion

13. #13. machine*

14. #14. pulsat*

15. #15. lifeport

16. #16. RM3

17. #17. preserv* OR stor*

18. #18. static

19. #19. university AND of AND wisconsin

20. #20. UW AND solution

21. #21. Marshall’s Soltran*

22. #22. Eurocollins

23. #23. HTK

24. #24. histidine AND tryptophan

25. #25. celsior

26. #26. viaspan

27. #27. soltran

28. #28. Belzer

29. #29. #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28

30. #30. #1 OR #10

31. #31. #29 AND #30

Health Technology Assessment (HTA) database on the CRD website

Search date: 29 November 2007

Search strategy same as for DARE

Additionally, the following databases of ongoing and recently completed trials were searched:

NRR (National Research Register)

2007, issue 4

Source: www.nrr.nhs.uk/

Search date: 21 November 2007

NB Includes information added until September 2007

ReFeR: Research Findings Register (now withdrawn)

Source: www.refer.nhs.uk/

Search date: 21 November 2007

Current Controlled Trials including MRC Trials database

Source: http://controlled-trials.com/

Search date: 20 November 2007

US Food and Drug Administration (FDA)

Source: www.fda.gov/

Search date: 5 May 2008

(a) Center for Drug evaluation and Research: Adverse Events reporting system.

(b) Center for Devices & Radiological Health

Medical Healthcare & Regulatory Authority

Source: www.mhra.gov.uk/

Search date: 5 May 2008

MEDLINE (1950 to date)

Dialog DataStar: online version

Search date: 8 February 2008

ECONOMICS#.W..DE.

HEALTH-CARE-ECONOMICS-AND-ORGANIZATIONS#.DE.

ECONOMICS-PHARMACEUTICAL#.DE.

ECONOMICS-NURSING#.DE.

ECONOMICS-MEDICAL#.DE.

ECONOMICS-HOSPITAL#.DE.

DIRECT-SERVICE-COSTS#.DE.

COST-OF-ILLNESS#.DE.

COSTS-AND-COST-ANALYSIS.DE.

COST-ALLOCATION.DE.

COST-BENEFIT-ANALYSIS.DE.

COST-CONTROL#.DE.

COST-OF-ILLNESS.DE.

COST-SHARING#.DE.

HEALTH-CARE-COSTS#.DE.

HEALTH-EXPENDITURES#.DE.

MODELS-ECONOMIC#.DE.

COST-SAVINGS.DE.

FEES-AND-CHARGES#.DE.

BUDGETS#.W..DE.

VALUE-OF-LIFE#.DE.

COST$3.TI,AB.

(ECONOMIC$2 OR PRICE$2 OR PRICING).TI,AB.

PHARMACOECONOMICS$OR PHARMA$3 ADJ ECONOMIC$

EXPENDITURE$2 NOT ENERGY

(EQ OR EUROQOL) ADJ (5D OR ‘5’ ADJ DIMENSIONS OR FIVE ADJ DIMENSIONS)

VALUE NEAR (MONEY OR MONETARY)

FISCAL OR FUNDING OR FINANCIAL OR FINANCE

(RESOURCE ADJ USE).TI,AB.

BUDGET.TI,AB.

EMBASE (1974 to date)

Dialog DataStar: online version

Search date: 8 February 2008

COST-EFFECTIVENESS-ANALYSIS#.DE.

COST-BENEFIT-ANALYSIS#.DE.

COST#.W..DE.

COST-CONTROL#.DE.

HOSPITAL-COST#.DE.

COST-MINIMIZATION-ANALYSIS#.DE.

COST-OF-ILLNESS#.DE.

COST–UTILITY-ANALYSIS#.DE.

DRUG-COST#.DE.

HEALTH-CARE-COST#.DE.

HEALTH-ECONOMICS#.DE.

ECONOMIC-EVALUATION#.DE.

PHARMACOECONOMICS#.W..DE.

ECONOMICS#.W..DE.

BUDGET.TI,AB.

BUDGET#.W..DE.

ECONOMIC-ASPECT#.DE.

FINANCIAL-MANAGEMENT#.DE.

HEALTH-CARE-FINANCING#.DE.

(PRICE$2 OR PRICING).TI,AB.

(FINANCIAL OR FINANC$3 OR FUNDING).TI,AB.

(FEE OR FEES).TI,AB.

(ECONOMIC$2 OR PHARMACOECONOMIC$2 OR PHARMACO ADJ ECONOMIC$2).TI,AB.

ECONOMIC$2.TI,AB.

COST$4.TI,AB.

NHS Economic Evaluation Database (NHS EED) on the CRD website

Search date: 8 February 2007

Same strategy as DARE databases (clinical effectiveness section above)

ISI Web of Knowledge (SCI-Expanded) (1970 to date)

Search date: 8 February 2008

TS=(economic* or price* or pricing or pharmacoeconomic* or pharma economic*)

TS=(cost* or budget)

TS=(value SAME (money or monetary))

The above were put together (OR) and combined (AND) with line #10 of the clinical effectiveness search.

MEDLINE (1950 to date)

Dialog DataStar: online version

Search date: 08 February,2007

QUALITY-OF-LIFE#.DE.

QUALITY-ADJUSTED-LIFE-YEARS#.DE.

VALUE-OF-LIFE#.DE.

(QUALITY ADJ ADJUSTED ADJ LIFE).TI,AB.

(QUALITY ADJ OF ADJ LIFE).TI,AB.

(QALY$2 OR QALD$2 OR QALE$2 OR QTIME$2).TI,AB.

(DISABILITY ADJ ADJUSTED ADJ LIFE ADJ YEARS).TI,AB. OR DALY$2.TI,AB.

HEALTH-STATUS-INDICATORS#.DE.

COST ADJ UTILITY

(SF36 OR SF ADJ ‘36’ OR SHORT ADJ FORM ADJ ‘36’ OR SHORTFORM ADJ ‘36’ OR SF ADJ THIRTYSIX OR SF ADJ THIRTY ADJ SIX OR SHORTFORM ADJ THIRTYSIX OR SHORTFORM ADJ THIRTY ADJ SIX OR SHORT ADJ FORM ADJ THIRTY ADJ SIX OR SHORT ADJ FORM ADJ THIRTYSIX OR SHORT ADJ FORM ADJ THIRTY ADJ SIX).TI,AB.

(SF6 OR SF ADJ ‘6’ OR SHORT ADJ FORM ADJ ‘6’ OR SHORTFORM ADJ ‘6’ OR SF ADJ SIX OR SFSIX OR SHORTFORM ADJ SIX OR SHORT ADJ FORM ADJ SIX).TI,AB.

(SF12 OR SF ADJ ‘12’ OR SHORT ADJ FORM ADJ ‘12’ OR SHORTFORM ADJ ‘12’ OR SF ADJ TWELVE OR SFTWELVE OR SHORTFORM ADJ TWELVE OR SHORT ADJ FORM ADJ TWELVE).TI,AB.

(SF16 OR SF ADJ ‘16’ OR SHORT ADJ FORM ADJ ‘16’ OR SHORTFORM ADJ ‘16’ OR SF ADJ SIXTEEN OR SFSIXTEEN OR SHORTFORM ADJ SIXTEEN OR SHORT ADJ FORM ADJ SIXTEEN).TI,AB.

(SF20 OR SF ADJ ‘20’ OR SHORT ADJ FORM ADJ ‘20’ OR SHORTFORM ADJ ‘20’ OR SF ADJ TWENTY OR SFTWENTY OR SHORTFORM ADJ TWENTY OR SHORT ADJ FORM ADJ TWENTY).TI,AB.

(EUROQOL OR EURO ADJ QOL OR EQ5D OR EQ ADJ 5D).TI,AB.

(HQL OR HQOL OR H ADJ QOL OR HRQOL OR HR ADJ QOL OR QOLY OR QOL).TI,AB.

(HYE OR HYES).TI,AB.

(HEALTH$2 ADJ YEAR$2 ADJ EQUIVALENT$2).TI,AB.

(HEALTH ADJ UTILIT$4 OR HUI OR HUI1 OR HUI2 OR HUI3 OR DISUTIL$6).TI,AB.

ROSSER.TI,AB.

(QUALITY ADJ OF ADJ WELL ADJ BEING).TI,AB. OR (QUALITY ADJ OF ADJ WELLBEING).TI,AB.

QWB.TI,AB.

(WILLINGNESS ADJ TO ADJ PAY).TI,AB.

(STANDARD ADJ GAMBLE$2).TI,AB.

(TIME ADJ TRADE ADJ OFF).TI,AB. OR (TIME ADJ TRADEOFF).TI,AB.

TTO.TI,AB. OR VAS.TI,AB.

(VISUAL ADJ (ANALOG OR ANALOGUE)).TI,AB.

(PATIENT ADJ PREFERENC$2).TI,AB

The above terms were put together with “OR” and combined (“AND”) with line 39 from the clinical effectiveness searches.

EMBASE (1974 to date)

Dialog DataStar: online version

Search date: 8 February 2008

QUALITY-OF-LIFE#.DE.

(QUALITY ADJ ADJUSTED ADJ LIFE).TI,AB.

SOCIOECONOMICS.W..DE.

(QALY$2 OR QALD$2 OR QALE$2 OR QTIME$2).TI,AB.

(DISABILITY ADJ ADJUSTED ADJ LIFE ADJ YEARS).TI,AB. OR DALY$2.TI,AB.

(SF36 OR SF ADJ ‘36’ OR SHORT ADJ FORM ADJ ‘36’ OR SHORTFORM ADJ ‘36’ OR SF ADJ THIRTYSIX OR SF ADJ THIRTY ADJ SIX OR SHORTFORM ADJ THIRTYSIX OR SHORTFORM ADJ THIRTY ADJ SIX OR SHORT ADJ FORM ADJ THIRTY ADJ SIX OR SHORT ADJ FORM ADJ THIRTYSIX OR SHORT ADJ FORM ADJ THIRTY ADJ SIX).TI,AB.

(SF6 OR SF ADJ ‘6’ OR SHORT ADJ FORM ADJ ‘6’ OR SHORTFORM ADJ ‘6’ OR SF ADJ SIX OR SFSIX OR SHORTFORM ADJ SIX OR SHORT ADJ FORM ADJ SIX).TI,AB.

(SF12 OR SF ADJ ‘12’ OR SHORT ADJ FORM ADJ ‘12’ OR SHORTFORM ADJ ‘12’ OR SF ADJ TWELVE OR SFTWELVE OR SHORTFORM ADJ TWELVE OR SHORT ADJ FORM ADJ TWELVE).TI,AB.

(SF16 OR SF ADJ ‘16’ OR SHORT ADJ FORM ADJ ‘16’ OR SHORTFORM ADJ ‘16’ OR SF ADJ SIXTEEN OR SFSIXTEEN OR SHORTFORM ADJ SIXTEEN OR SHORT ADJ FORM ADJ SIXTEEN).TI,AB.

(SF20 OR SF ADJ ‘20’ OR SHORT ADJ FORM ADJ ‘20’ OR SHORTFORM ADJ ‘20’ OR SF ADJ TWENTY OR SFTWENTY OR SHORTFORM ADJ TWENTY OR SHORT ADJ FORM ADJ TWENTY).TI,AB.

(EUROQOL OR EURO ADJ QOL OR EQ5D OR EQ ADJ 5D).TI,AB.

(HQL OR HQOL OR H ADJ QOL OR HRQOL OR HR ADJ QOL OR QOLY OR QOL).TI,AB.

(HYE OR HYES).TI,AB. OR (HEALTH$2 ADJ YEAR$2 ADJ EQUIVALENT$2).TI,AB.

(HEALTH ADJ UTILIT$4 OR HUI OR HUI1 OR HUI2 OR HUI3 OR DISUTIL$6).TI,AB.

ROSSER.TI,AB.

(QUALITY ADJ OF ADJ WELL ADJ BEING).TI,AB. OR (QUALITY ADJ OF ADJ WELLBEING).TI,AB. OR QWB.TI,AB.

(WILLINGNESS ADJ TO ADJ PAY).TI,AB.

(STANDARD ADJ GAMBLE$2).TI,AB.

(TIME ADJ TRADE ADJ OFF).TI,AB. OR (TIME ADJ TRADEOFF).TI,AB.

TTO.TI,AB. OR VAS.TI,AB.

(VISUAL ADJ (ANALOG OR ANALOGUE)).TI,AB.

(PATIENT ADJ PREFERENC$2).TI,AB.