E‑vita open plus for treating complex aneurysms and dissections of the thoracic aorta

Neither the sponsor nor the external assessment centre identified any relevant published economic evidence for the E‑vita open plus or for the comparator techniques.

The sponsor submitted a de novo analysis comparing the use of the E‑vita open plus against a 2-stage classical 'elephant trunk' procedure in terms of overall costs, in-hospital mortality and adoption rates. The population was a cohort of 3,500 people with aneurysms, dissections and other specified lesions of the thoracic aorta. The model consisted of 2 decision trees over a 1-year time horizon:

• a current practice model using the classical 'elephant trunk' procedure

• an intervention model comparing current practice against use of the E‑vita open plus at a 40% adoption rate.

The first stage of the current practice arm was divided into 2 options: woven graft or branched graft. For patients undergoing stage 2, the options were woven graft (open surgery) or endovascular stent graft. Following clarification from the sponsor, these options were further defined as follows:

• Woven graft at stages 1 and 2 referred to 2-stage open surgical repair with vascular graft placement.

• Woven graft followed by stent graft at stage 2 referred to 2-stage repair with open surgical graft placement in the ascending aorta and arch and endovascular stent graft placement in the descending aorta.

• Branched graft followed by woven or stent graft at stage 2 referred to open surgical 'debranching' of the head and neck vessels with endoluminal stent graft placement in the aortic arch and either a vascular graft or endovascular stent graft in the descending aorta.

The sponsor carried out 1- and 2-way sensitivity analyses to include base-case, worst-case and best-case scenarios in its analysis.

The key assumptions used in the model were:

• In-hospital and 30‑day mortality was 15% for the E‑vita open plus, based on data from the paper by Jakob et al. (2011).

• The remaining 85% of patients treated would not experience other complications.

• Mortality rates were 20% for woven graft and 30% for branched graft at stage 2.

• The number of inpatient days for the classical elephant trunk procedure was 10 at stage 1 and 15 at stage 2.

• The number of inpatient days for the endovascular stent procedure was 10 at stage 1 and 8 at stage 2.

• The number of inpatient days for the E‑vita open plus was 4 in the intensive care unit and 6 in a surgical ward.

Technology costs were provided by the sponsor. The cost of the E‑vita open plus was £10,500 and the comparator costs were £200 for a woven graft for stages 1 and 2, £1,000 for a branched graft, and £5,000 for an endovascular stent graft. Consumable costs (mainly for the guide wire, estimated to be £130) were not included because the sponsor considered these to be the same for the technology and comparators. Cost estimates for clinical time and resource use were sourced from published literature, the Personal Social Services Research Unit (PSSRU) unit costs of health and social care manual, and NHS reference costs.

The cost of a surgeon was estimated to be £399 per hour and the costs of a perfusionist and anaesthetist were each estimated to be £87 per hour (registrar rate). The cost for theatre time, including nursing and consumables, was estimated to be £24 per hour, and £30 per hour for corresponding intensive care unit costs. These were derived from the NHS tariff for admitted patient case and outpatient procedures but no codes were specified. The sponsor used a daily cost for an intensive care unit stay of £1,500, taken from a report in The Lancet. The cost of a surgical ward inpatient stay was taken to be £420 per day, based on 2 different tariff codes. The cost of death cited by the sponsor (£8,000) was taken from a cancer network publication.

The sponsor carried out 1-way sensitivity analyses varying the adoption rate from 20% to 100% (in the modelled population of 3,500 patients assumed to be eligible for the E‑vita open plus). The proportion of woven or branched grafts used at stage 1 was varied from 60% to 95% from a base-case estimate of 85%. The proportional suitability for a second stage operation was varied from a base case of 80% to 60% and 95%, and the proportion of patients having each stage 2 procedure was varied from a 50% base case to 40% and 100%. The sponsor also carried out a 2-way sensitivity analysis varying the in-hospital death rate at stage 1 of the classical elephant trunk procedure and for the E‑vita open plus.

The sponsor presented the results of its de novo analysis as an average cost per patient, assuming a 100% adoption level (rather than the 40% adoption level described in the model decision tree in section 5.2) for the E‑vita open plus compared with current practice. The cost for the E‑vita open plus was £25,689 and the overall cost presented for the comparators was £30,241, indicating a cost saving per patient of £4,553 if the E‑vita open plus was used. The cost per patient varied across the different comparators, ranging from £26,691 for woven graft (stage 1) with endovascular stent (stage 2) to £36,016 for branched graft (stage 1) with woven graft (stage 2).

The sponsor's sensitivity analysis showed little variation in the cost savings generated for the E‑vita open plus at different adoption levels, with an average cost saving of around £4,358. The sponsor reported that varying the parameters for second-stage suitability and in-hospital death had an impact on the cost savings, but that this was relatively small. Varying the suitability of a second stage operation produced higher cost savings per patient if the level of suitability was raised. The sponsor did not consider the subgroups defined in the scope in its de novo analysis because of a lack of available data on the comparators.

The external assessment centre considered that the sponsor's de novo cost model was flawed because it did not include the short- or long-term costs of complications and because some of the costs and clinical parameters were inaccurate or inappropriate. Specifically, the external assessment centre considered that a per-patient, rather than a cohort, approach would have been more useful and that the estimated cohort of 3,500 patients was too large.

The external assessment centre carried out additional modelling to address these issues, constructing short- and long-term models that included the costs of complications (stroke, paraplegia, renal failure and bleeding). Both models compared per-patient costs for the E‑vita open plus and the 3 comparators defined in the scope. A decision tree was constructed for each procedure in each model. In-hospital mortality was modelled at each stage of each procedure. The time horizon for the short-term model was 1 year, as the external assessment centre considered that stage 2 procedures were likely to be carried out within 6 months of stage 1. The long-term model had a 20 year time horizon, based on the UK life expectancy of the average age (65 years) of those receiving treatment described in published literature. Lifetime costs of complications were included in this model.

In the short-term model for each comparator, patients with no complications or bleeding at stage 1 were assumed to proceed to the second stage procedure, whereas it was assumed that those who had a stroke, renal failure or paraplegia would not. In the long-term model, the annual costs of care for stroke, paraplegia and renal failure were taken from published literature and discounted at 3.5%. The discounted annual cost was multiplied by a survival probability for 65 to 85 years and the weighted annual costs were summed to estimate the lifetime cost of the complications.

The external assessment centre estimated the probabilities of the outcomes at each stage from the register data for the E‑vita open plus and from its meta-analysis of the clinical evidence on the comparator procedures. The probability of paraplegia at stage 1 was assumed to be the same for 2-stage repair with vascular graft and 2-stage repair with endovascular stent graft. For open debranching, the probability of paraplegia and that of renal failure at stage 1 was taken from hybrid procedure estimates. Operating times and total lengths of stay for all the comparators were sourced from published literature. Operating time for the E‑vita open plus and details of the surgical team involved for each procedure were taken from the sponsor's model. The team included a consultant surgeon, consultant anaesthetist, associate specialist, perfusionist and 2 specialist nurses. A consultant radiologist was included for stage 2 procedures involving stent grafts.

The external assessment centre used technology and comparator costs from the sponsor's model but derived more precise estimates for staff and ward stay costs than those used by the sponsor (as described in section 5.7). Costs for each professional in the surgical team were taken from PSSRU 2012 unit costs. The costs for an intensive care unit and surgical ward stay were sourced from NHS reference costs at £1,410 per day and £383 per day respectively.

Complications were assumed to incur additional in-hospital management costs and a single cost figure was applied across all procedures (£2,155). The annual cost for stroke care was estimated to be £9,597 at 2012 prices, from the NICE guideline on atrial fibrillation: diagnosis and management. The annual cost of paraplegia was estimated to be £14,580, based on published literature and inflated to 2012 prices. The annual cost for renal failure used was £32,961, taken from the NICE guideline on renal replacement therapy and conservative management and inflated to 2012 prices.

Costs for multiple stents were included in the analysis and sourced from the sponsor's submission. However, the external assessment centre was not able to model the probable outcomes from using multiple stents, citing a lack of available clinical evidence.

Results of the short-term model showed that treatment with the E‑vita open plus could generate a cost saving of £280 per patient when compared against 2-stage repair with vascular graft. The E‑vita open plus incurred costs when compared against 2-stage repair with endovascular stent graft (£4,760) and also when compared against open debranching with endoluminal stent graft (£7,663).

Results from the long-term model, which included the lifetime costs of complications, showed that treatment with the E‑vita open plus could generate significant cost savings when compared with all 3 comparator procedures. The estimated saving per patient 20 years after the procedure was £41,213 when compared against 2-stage repair with vascular graft, £39,392 when compared against 2-stage repair with endovascular stent graft and £51,778 when compared against open debranching with endoluminal stent graft.

The external assessment centre carried out a deterministic sensitivity analysis to investigate the impact of uncertainty on the likelihood and costs of complications. The probabilities of in-hospital mortality and paraplegia were varied separately based on their 95% confidence intervals from the external assessment centre's meta-analysis of the clinical evidence. The overall management costs and the annual costs of complications were varied separately using minimum and maximum ranges identified in the literature, if available. The proportion of days spent in an intensive care unit and the cost of an intensive care unit day were also varied to reflect uncertainty in the number of organs needing support. Results of the sensitivity analysis showed that varying the probability of in-hospital mortality and paraplegia for the E‑vita open plus, or the management or annual costs of complications, did not substantially change the expected cost savings in the base-case estimate. In the short-term model, varying the proportion of days spent in the intensive care unit did change the observed cost savings. At a 20% level, the E‑vita open plus incurred costs when compared with all 3 comparators. At a 60% level, there were cost savings for the E‑vita open plus compared against 2-stage repair with vascular graft. Varying the cost of intensive care unit stay affected the short-term model results in a similar way, but neither variable substantially altered the cost savings in the long term.

The external assessment centre acknowledged some limitations in its model. Complications were assumed to occur only in the short term, because data were not available about their occurrence in the longer term. However, clinical expert advice indicated that the majority of complications would occur during or shortly after the intervention. The external assessment centre recognised that a more complex model (such as Markov or discrete event simulation) might have facilitated a more refined analysis and noted that the sensitivity analysis did not account for the possibility of multiple complications occurring in individual patients.

The external assessment centre developed a profile of year-on-year costs for the long-term model from year 1 to year 20. The E‑vita open plus was estimated to be cost saving at and beyond 2 years after the procedure compared with all 3 comparator procedures. For example, when compared with 2-stage repair with vascular grafting, the base-case cost savings per person treated with the E‑vita open plus were £6,057 at 2 years after the procedure, £13,822 at 5 years, and £24,948 at 10 years. When compared with 2-stage repair with endovascular stent grafting, the cost savings were £1,471 at 2 years after the procedure, £9,847 at 5 years, and £21,847 at 10 years. When compared with open surgical debranching with endoluminal stent grafting, the cost savings were £726 at 2 years after the procedure, £12,003 at 5 years, and £28,158 at 10 years.

For the guidance review, the external assessment centre revised the model to reflect 2018 costs. The major changes in the update relate to acute care costs of adverse events and staff costs. In the original model, the acute care cost of adverse events was calculated as £2,155; in the revised model, the costs depend on the type of adverse event and range from £498 for bleeding to £11,663 for paraplegia. Staff costs in the revised model were taken from the Personal Social Services Research Unit (PPSRU) 2017 and were often cheaper. Base-case results for the 2018 revised model show that estimated cost savings per patient at 5 years after the procedure are:

• £13,334 compared with 2-stage repair involving open insertion of a vascular graft.

• £10,225 compared with 2-stage repair involving endovascular stent grafting.

• £12,536 compared with open surgical debranching followed by endoluminal stent grafting.
  
  These saving increase across the 3 comparators in the longer term. Further details of the 2018 revised model are in the revised model summary. [2018]