Many studies were based on randomised controlled clinical trials. In three of the studies considered here 567, becaplermin efficacy was derived from a meta-analysis conducted by Smiell et al. 8 of four randomised studies including 922 patients with diabetic foot ulcers. Kantor and Margolis 9 took the becaplermin efficacy data from a review by Wieman 10 of four multicentre, randomised, placebo-controlled, parallel group studies. These four studies included a total of 922 patients. Sibbald et al. 11 used the multicentre double-blind placebo-controlled phase III trial of 382 patients by Wieman et al. 12. In the Apligraf study by AÉTMIS 13, data were derived from the pivotal clinical trial by Falanga et al. 14 including 293 venous ulcer patients to calculate the primary outcome measure. Steinberg et al. 15 took effectiveness data from the randomised prospective trial by Veves et al. 16 to evaluate the cost-effectiveness of Apligraf in the treatment of diabetic foot ulcers. Finally, in the Dermagraft study by Segal and John 17, the US pivotal clinical trial by Naughton and colleagues 18 of 235 patients was used to assess the mean cost per ulcer healed.

An overview of the sources for effectiveness data used in the evaluations is given in table S5 [see Additional file 5]. Table 2 provides an overview of the studies and systematic reviews used for the assessment of effectiveness.

As the identified health economic evaluations were too heterogeneous and did not report adequate details for a meta-analysis of results, we performed a narrative synthesis of economic evidence including a description of all relevant information regarding study methods and cost-effectiveness results. In addition, in table 3, a summary table is provided presenting key information relating to type of ulcers, country, perspective, comparison of interventions, primary measure of health benefit and incremental or average cost-effectiveness ratios.

In our literature search, we identified 5 health economic evaluations that assessed the cost-effectiveness of Apligraf^® in the management of venous leg or diabetic foot ulcers. All but one study were cost-effectiveness models and most economic evaluations were funded by manufactures/patentees of the products. Due to methodological weaknesses (i.e. only average cost-effectiveness ratios, small sample sizes, short time horizons, inadequate treatment of uncertainty), the quality of the evidence was considered limited.

An analytical prediction model was developed at AÉTMIS (2001) 13 by assembling information for the following treatment options of venous leg ulcer patients: compression alone, compression and Apligraf simultaneously, compression and Apligraf for hard-to-heal ulcers which are unresponsive to conventional therapy. To assess the number of ulcer days averted, which was used as the benefit measure in the economic analysis, data from the pivotal study by Falanga et al. (1998) 14 were obtained (further information on clinical trial characteristics are available in table 2). From both a health-care system perspective and a societal perspective the incremental cost for each ulcer day averted was calculated to amount to Can $26 when compression and Apligraf were used simultaneously, and Can $22 when compression and Apligraf were used for hard-to-heal ulcers.

Harding et al. (2000) 19 combined a summary of published studies and expert opinion to compare the cost-effectiveness of saline gauze, the hydrocolloid dressing Granuflex^® (UK trade name for DuoDERM^®), and the human skin equivalent Apligraf for venous leg ulcer patients, over a 12-week treatment period. They identified 12 studies involving 843 ulcers, of which 205 were treated with saline gauze, 509 with Granuflex, and 278 with Apligraf. Due to heterogeneity across studies, only the percentage of ulcers healed was reported and used for the effectiveness estimate. The views of a European panel of four wound-care specialists on the issue of resource utilisation and costs were used to design protocols of care where data were not available from the literature. No summary measure of benefit was provided in the economic analysis. Cost-effectiveness was calculated from a health-care payer's perspective as the total medical cost of care of the cohort divided by the number of healed wounds, which yielded a cost per healed wound of £342 for Granuflex (nurse time £97, compression £121, dressing £124 and other £0), £541 for saline gauze (nurse time £327, compression £166, dressing £48 and other £0), and £6,741 for Apligraf (nurse time £70, compression £144, dressing £6,526 and other £1). An incremental cost-effectiveness ratio was not calculated.

Kerstein et al. (2001) 20 undertook a study in the United States, similar to that by Harding et al. 19, with a broader literature review to assess the cost-effectiveness of gauze dressings impregnated with saline, paraffin or zinc oxide, as compared with hydrocolloid dressings, and human skin construct (Apligraf) in venous leg ulcer patients. They included 18 studies and identified 223 patients on impregnated gauze dressings, 530 on hydrocolloid dressings, and 130 on Apligraf. The data on effectiveness were derived from a review of published studies. The benefit measure used in the economic analysis was the number of persons healed or not healed in a hypothetical managed-care plan with 100,000 covered lives. The average cost per patient healed was lowest with the hydrocolloid dressing DuoDERM (US $1,873), followed by impregnated gauze dressings at US $2,939, and then the human skin equivalent Apligraf at US $15,053 over a treatment period of 12 weeks.

Meaume and Gemmen (2002) 21 also used a similar methodology to that adopted by Harding et al. 19 to evaluate the cost-effectiveness of saline gauze, the hydrocolloid dressing DuoDERM (German trade name Varihesive^® and UK trade name Granuflex), and Apligraf in venous leg ulcer patients from a European and French perspective, respectively. The French perspective in this study was provided by a panel of five French wound-care experts. The European expert panel comprised four specialists from the UK, France, Germany, and Sweden. The evidence on effectiveness was derived from a literature review. The benefit measure used in the economic analysis was the number of patients healed at 6 and/or 12 weeks in a hypothetical cohort of 100,000. Although the treatment patterns are different between the United Kingdom and France, the pattern of cost-effectiveness per patient healed was consistent with the study conducted by Harding et al. 19 with the hydrocolloid dressing DuoDERM being most cost-effective at £2,763/1,018, followed by saline gauze at £1,436/1,722 and Apligraf at £11,396/15,920 from a European and a French perspective, respectively.

In the economic analysis by Steinberg et al. (2002) 15 the living skin equivalent Apligraf was compared to saline-moistened gauze to assess the cost-effectiveness of these two dressings in diabetic foot ulcer patients, on the basis of data from a six-month, multicentre, randomised trial 16. In this trial, there were 112 patients in the intervention group and 96 controls. The benefit measures used in the economic analysis were the number of ulcer-free months gained and the number of amputations or resections avoided. In comparison to controls, patients in the living skin equivalent group had a higher average number of ulcer-free months (2.3 in the intervention group vs. 1.5 in the control group) and a lower average number of amputations or resections (5.4% in the intervention group vs. 12.5% in the control group). With a follow-up of six months, the incremental cost-effectiveness ratio of Apligraf over saline-moistened gauze was US $6,683 when the benefit measure was based on the number of ulcer-free months gained, and US $86,226 when amputations or resections avoided were considered.

Segal and John (2002) 17 used observational case studies and a Markov model to evaluate the cost-effectiveness of Dermagraft in addition to conventional management in diabetic patients with neuropathic foot ulcers. The Markov model was used to assess average cost per ulcer healed, on the basis of effectiveness data from the pivotal clinical trial 18. The benefit measures used in the model were the incremental healed weeks. On the basis of an observational case study of 27 hard-to-heal ulcers an alternative cost-effectiveness estimate was established, using a prospective analysis of resource use and cost for Dermagraft. A survey in the specialist clinic setting was used to assess the average cost of conventional management. On the basis of the Markov model, the average cost per ulcer healed was A $10,906 with conventional management and A $12,128 with Dermagraft as an additional treatment option. Using the results of the observational case studies, the average cost to treat an ulcer was A $12,500 prior to Dermagraft treatment and A $4,682 after starting Dermagraft treatment respectively.

In our literature search, we found 5 economic evaluation studies that assessed the cost-effectiveness of becaplermin in the management of diabetic foot ulcers. All studies were cost-effectiveness models. Most of these cost-effectiveness analyses were funded by pharmaceutical companies or coauthored by their employees, and reported results that favoured the sponsor's product. The quality of the evidence was considered high.

The following three studies 567 used the same source of effectiveness evidence. However, country-specific patterns of resource usage and prices have a high impact on cost-effectiveness estimates. Therefore, to ensure that country-specific differences in treatment patterns and cost estimates are appropriately accounted for, the findings of these studies are reported here separately.

Persson et al. (2000) 5 used a Markov model to evaluate the cost-effectiveness of using becaplermin gel in combination with good wound care, as opposed to good wound care alone, to treat diabetic patients in Sweden who presented neuropathic, lower extremity ulcers. As part of a recent study on the cost of illness, a Markov model was developed 22. With that as a starting point, the authors developed a more complete model of diabetic lower extremity ulcers. Data on resource usage were derived from a series of Swedish studies 23242526. The becaplermin efficacy was taken from a combined analysis of four randomised studies 8. The primary benefit measure used in the model was the number of ulcer-days averted. In comparison to controls, patients in the intervention group had a higher average number of ulcer-free months (4.22 in the intervention group vs. 3.41 in the control group) and a lower average number of amputations (5.91% in the intervention group vs. 6.50% in the control group). The average expected costs were US $12,078 for good wound care alone and US $11,708 for becaplermin.

In the economic analysis by Ghatnekar et al. (2000) 6 a similar Markov model to that developed by Persson et al. 5 was used to evaluate the cost-effectiveness in the UK of becaplermin gel as an adjunct therapy to good wound care in diabetic foot ulcer patients. Data on resource usage were taken primarily from a Swedish study 25. To ensure the generalisability of the study results to the UK setting, they were reviewed by an expert panel of UK physicians. Data on becaplermin efficacy were drawn from the analysis by Persson et al 5. The primary benefit measure used in the model was the number of ulcer-days averted. As in the analysis by Persson et al. 5 the average number of months spent in the healed state rose by 24% from 3.41 to 4.22 and the average number of amputations fell from 6.50% to 5.91% – a decline of 9%. The average expected costs over a 12-month time horizon were £10,880 for good wound care alone and £10,403 for treatment with becaplermin.

Ghatnekar et al. (2001) 7 used the same Markov model as the one developed by Persson et al. 5 to assess cost-effectiveness in four European countries (France, Sweden, Switzerland, UK) of becaplermin as an adjunct treatment option to good wound care alone in patients with diabetic foot ulcers. Another objective was to address the issue of generalisability by analysing the effect of different resource use patterns on the economics of managing diabetic foot ulcers. Data on effectiveness evidence for becaplermin were drawn from a pooled analysis of four randomised trials 8. Resource data were primarily taken from a series of Swedish studies 232425. The primary benefit measure used in the model was the number of ulcer-months avoided. The expected number of months in the healed state was higher in the becaplermin group, 4.22 months vs. 3.41 months – an increase by 24%. Furthermore it was assumed that the average number of amputations would decline by 9%, from 6.50 to 5.91 per 100 patients. The average expected costs over a 12-month time horizon in the becaplermin group and the control group respectively were US $11,977/11,993 for France, US $12,168/11,783 for Sweden, US $14,112/13,832 for Switzerland, and US $17,601/17,133 for the UK.

Kantor and Margolis (2001) 9 used published literature and a database to assess the cost-effectiveness of various treatment options for diabetic neuropathic foot ulcers. The following treatment strategies were considered: standard care, standard treatment in a wound care centre, becaplermin, or platelet releasate. The becaplermin data were taken from a meta-analysis conducted by the authors on published studies of becaplermin gel 0.01% 10. The measure of benefit used in the economic analysis was the percentage of ulcers healed. At 20 weeks of care the percentages of ulcers healed were 30.9% for standard care, 43% for becaplermin, 36.8% for platelet releasate, and 35.6% for treatment in a wound care centre. The incremental cost per patient of increasing by 1% the chance of healing at 20 weeks was US $36.6 for standard care vs. becaplermin, and US $70.86 for becaplermin vs. treatment in a specialised wound care centre. At this time, becaplermin was superior to platelet releasate, as a more effective and less costly treatment.

In the economic analysis by Sibbald et al. (2003) 11 a decision tree was used to assess the cost-effectiveness of becaplermin as an adjunct treatment option available to best clinical care in patients with nonhealing neuropathic diabetic foot ulcers. The effectiveness data were taken from a randomised controlled clinical trial of 251 diabetic patients 12. The benefit measure used in the economic analysis was the number of ulcer-days averted. The number of ulcer-days per patient was 237 in the intervention group and 211 in the control group over a one year time horizon. Based on the model, the incremental cost-effectiveness ratio was Can $6 per ulcer-day averted.

While the cost-effectiveness of tissue engineered skin was considered poor in three studies 192021, one study found it to be more cost-effective than compression alone 13. This difference could be explained by the assumptions made in each analysis, most importantly the frequency of tissue engineered skin application in each model. Table 3 provides an overview of the economic evaluations. As described in the methods section, the last column is based on an appraisal of studies by two independent reviewers according to two published quality checklists.

Ho et al. 2 included in their review three cost-effectiveness analyses of Apligraf 272829, and one cost-effectiveness study of Dermagraft 1. Studies assessed by that review were excluded from the present analysis to avoid 'double counting' of evidence, thus enhancing the efficient use of resources. A summary can be found in Table 4.

In the four health economic evaluations, univariate and multivariate sensitivity analyses respectively were conducted on probabilities and costs. The results of the sensitivity analyses showed that the models were generally robust to almost all variations in model inputs. The study results by Sibbald et al. 29 were sensitive to changes in the time loss from usual daily activities, in which case Apligraf plus four-layer bandage became the dominant treatment modality. In the analysis by Schonfeld et al. 28, when the cost of hospitalisation for patients in the Apligraf group was doubled, Apligraf was no longer the dominant strategy. In the economic evaluation by Redekop et al. 27, the incremental cost-effectiveness ratio was seen to be sensitive to the required number of applications of Apligraf, to amputation costs and differences in infection rate. For all three parameters the results of the incremental cost-effectiveness ratios ranged from cost-effective to cost-saving.

In none of the studies the clinical evidence was solely based on the 'gold standard' of randomised, clinical trials. Instead, cost-effectiveness was frequently modelled on the basis of observational data and, in some cases, supplemented by the author's assumptions and/or expert opinions.

All studies used condition-specific measures of benefit: percentage of ulcers healed 9192021, ulcer-days averted 61113, additional healed weeks 17, ulcer-free months gained 5715 and amputations avoided 15. While these may represent standard effectiveness measures of wound care, these cost-effectiveness ratios do not allow for a meaningful comparison of cost-effectiveness across disease areas or a comparison with societal willingness to pay for a health gain. A more generic measure, for example the quality-adjusted life-years, would be desirable. Various studies considered health-related issues on overall quality of life in patients with venous leg and diabetic foot ulcers respectively 30313233, but our literature research identified no cost-utility analyses in this field. This may be due to a lack of studies that take into account quality of life in the health care interventions considered here. Only one study could be identified, in which quality of life with relation to the use of Apligraf was taken into account 34.

An important limitation of the publications considered in this review was the short time horizon. In three studies 192021, the time horizon was twelve weeks, which gives rise to the following problem 20: as a result of the short time horizon, the average healing cost is comparatively high, because dressings are applied in the first weeks of treatment. Thus, over a longer time horizon, the average treatment cost may be expected to become lower. In some studies, the time horizon was limited to that used in the primary studies. Two studies considered longer time periods in one-way sensitivity analyses 56. The authors of these studies suggest that the cost-effectiveness may be more favourable over a longer time horizon.

In some studies, the results were reported as cost per wound healed. This does not consider additional factors, such as the length of time needed for the wound to heal, pain, or other complications which may arise during the healing process. It may be that a particular treatment has a higher cost per wound healed compared to other treatments, but helps heal a larger number of wounds.

None of the identified products fully met all quality criteria and frequently deviated from current standards of health economic evaluation. For example, Harding et al. 19, Kerstein et al. 20, and Meaume and Gemmen 21 did not calculate incremental cost-effectiveness ratios but only reported average cost-effectiveness. In many cases, the perspective was not stated clearly or substantiated. Also, quantities of consumed resources or details on discounting were often not given. Probabilistic sensitivity analysis was conducted in none of the eleven health economic evaluations. Most of the economic evaluations relied on one-way sensitivity analyses.

The categories of costs included in the evaluations differed substantially between studies. Most studies considered only direct costs (e.g. outpatient clinic visits, hospital days, physician office visits, homecare, laboratory and diagnostic tests, debridement, dressing materials, orthopaedic appliances, costs of amputations). Indirect costs were included in two evaluations only 1113. Intangible costs were not evaluated. In many studies, a detailed breakdown of the costs was not provided, the price year was not reported, and the unit costs were not presented separately from the quantities of resources used. Health resource use was often estimated on the basis of expert opinions.

The transferability of clinical and economic evidence is an important issue in health economic evaluation 35. Some authors noted that a limitation of their studies was the lack of available data. In most evaluations, effectiveness data came from clinical trials conducted in the USA, but the application of US data to European countries could be controversial. Furthermore, the number of applications of skin substitutes may vary substantially between settings. The issue of generalisability of study results to other settings was extensively addressed only by one study 7, in which four different European countries were considered. The study by Ghatnekar et al. (2001) revealed substantial differences in treatment patterns and in the nature of cost-savings across countries, owing to the differences in the health care systems. In this context, reimbursement systems, relative prices, and treatment practices are important issues that vary from country to country 5. In general, decision-makers in other health care systems should investigate the cost-effectiveness of tissue-engineered products in their own setting.

Discounting was of low importance since all costs were incurred within a period of less than a year. However, while many guidelines recommend equal discounting for costs and effects, discounting of costs only was performed in two studies 56.

First of all, it must be stated that the choice of dressings studied was dictated by the available fully published health economic evaluations. Currently, only few chronic leg ulcer treatment modalities provide sufficient details from which to draw conclusions concerning cost-effectiveness. Furthermore, only studies in English, French or German could be included. The economic analyses were appraised by a checklist on methodological quality and the studies considered were rated only on the basis of widely applicable criteria known to the authors. However, good health economic evaluations are more than the sum of their parts 36. Even though the study appraisal was based on the consensus of two independent reviewers, it is nonetheless restricted by those reviewers' subjectivity.