5.1 The company identified 7 studies that incorporated a cost-effectiveness analysis. It did not rely on these economic studies for its model, but the structure of the de novo model is similar to that described in Gadler et al. (2016). The external assessment centre (EAC) judged the company's search strategy and inclusion/exclusion criteria reasonable, but noted that it could be improved with access to more databases and a more thorough strategy. The EAC considered that the population used by the company in its selection of economic evidence – 'patients implanted with cardiac resynchronisation therapy-defibrillators (CRT‑Ds)' – differed from the population specified in the scope. The company's population is broader and the EAC acknowledged that this probably reflects the lack of detail in the published evidence on the specific criteria used to define heart failure and CRT‑D use from the NICE technology appraisal guidance on implantable cardioverter defibrillators and cardiac resynchronisation therapy for arrhythmias and heart failure.
5.2 The EAC excluded 3 studies included by the company because they were outside the scope. Boriani et al. (2013) report on a model comparing hypothesised CRT‑Ds with 4-year and 7-year lifespans over a 15-year time horizon. The devices were not specifically named technologies and the lifespans were not based on data, but were chosen specifically to investigate how battery life affects costs. Biffi et al. (2011) focused on implantable cardioverter defibrillators and included only 10 patients with CRT‑Ds. It did not include devices from Boston Scientific. The Chung et al. (2015) abstract does not directly compare specific devices although it includes a device survival curve based on manufacturer data, but looks at the costs for different patient groups using devices with different lifespans.
5.3 Gadler et al. (2016) describes an economic model with a 6-year time horizon based on the data from Landolina (2015; see section 3.5), Swedish ICD and Pacemaker Registry and Swedish public tendering data. Survival data were taken from Yao et al. (2007). The authors found in the base case that using Boston Scientific devices reduced replacement procedures and saved SEK 19,172 (£1,687 based on XE.com currency conversion on 15 July 2016) per patient over 6 years. The study was funded by Boston Scientific.
5.4 Landolina et al. (2016) is an economic analysis based on a subset of the data from Landolina et al. (2015), with a 6-year time horizon and 2 perspectives: a hospital perspective and the Italian healthcare system perspective. Boston Scientific provided funding for the economic analysis. Of 1,726 heart failure patients in Landolina 2015, 1,399 were included in the economic analysis. The analysis compared devices released between 2007 and 2010 with devices released between 2003 and 2007 for 3 manufacturers (Boston Scientific, Medtronic and St Jude Medical) and for all manufacturers together. Weighted average prices of the devices were taken from tender information. The authors found that among recent-generation CRT‑Ds from different manufacturers, the total cost per patient over 6 years ranged from €25,579 to €31,536 (£21,665 to £26,711 based on XE.com currency conversion on 12 July 2016), with a maximum difference in cost of 40% for hospitals and 19% for the Italian healthcare system.
5.5 Priest et al. (2015) is a published abstract from a conference poster presentation comparing the costs for industry-standard and longer-lifespan devices from an Australian health system perspective over 15 years, using real-world data for implantable cardioverter defibrillators and CRT‑Ds (using the methods described by Boriani et al. 2013). Patient survival following first implant was taken from published literature. Average battery life was taken from a recent NICE review (not specified, but the figures quoted are found in NICE's technology appraisal guidance on implantable cardioverter defibrillators and cardiac resynchronisation therapy for arrhythmias and heart failure), and Boston Scientific real-world battery life data from more than 100,000 for implantable cardioverter defibrillators using the LATITUDE NXT remote monitoring system. The study concluded that if all patients switched from industry-standard devices to longer-lifespan batteries, the average cost per patient would fall by 19% and overall number of replacements would fall by 70%. This would result in cumulative cost savings of more than $900 million over 15 years.
5.6 The paper by Duxbury et al. (2014) is a published abstract from a conference poster presentation reporting the economic impact of implanting cardiac devices with longer lifespans from a UK perspective. The methodology was similar to that of the Priest et al. study (2015), in that it was based on Boriani et al. (2013). It also used the average lifespans described in the NICE technology appraisal and Boston Scientific real-world battery life data using the LATITUDE NXT remote monitoring system. The authors modelled the potential cumulative costs over 10 years for industry-standard and longer-lifespan devices using real-world battery data for implantable cardioverter defibrillators and CRT‑Ds. The study concluded that using devices with longer battery life could result in cumulative savings of up to £158 million over 10 years.
5.7 The EAC identified that the main weakness of the published economic evidence was it relates to devices no longer marketed, because of the rapid turnover of new models of the technology. The study by Gadler et al. (2016) was funded by Boston Scientific, so may be subject to bias. The EAC considered the lifespan data for the LATITUDE NXT system used in Priest et al. (2015) may not be directly comparable with that reported in NICE's technology appraisal guidance on implantable cardioverter defibrillators and cardiac resynchronisation therapy for arrhythmias and heart failure, because the patient populations may be different. The EAC concluded that because the Priest et al. (2015) and Duxbury et al. (2014) studies are only available as abstracts, the results should be treated with caution.
5.8 The company presented a de novo economic model adapted from Gadler et al. (2016) estimating mean cost savings per patient. The model is a decision tree with a 6-year time horizon and an NHS perspective. It compares Boston Scientific ENDURALIFE‑powered CRT‑Ds with Medtronic and St Jude Medical CRT‑Ds. For each device there are branches for procedural complications or no complications, with further branches for death, replacement or no replacement at 1 year and at each subsequent year. Clinical data in the model are taken from the Landolina et al. (2016) study on event-free battery survival and Yao et al. (2007) for cumulative probability of patient survival. The incidence of complications is taken from Tang et al. (2010) and the follow-up arrangements from NHS England 2013/14 NHS standard contract for cardiology: implantable cardioverter defibrillator and cardiac resynchronisation therapy (adult). The model assumes follow-up appointments at 6-month intervals with an additional post-procedure appointment.
5.9 The company's scenario analyses included exploring differences in device survival and device cost to identify thresholds at which the model becomes cost neutral. The price was varied by ±20% for each device separately using a one-way sensitivity analysis. The analyses showed that the cost model is highly sensitive to changes in both device survival and device cost. Higher device survival resulted in a marked decrease in relative costs. The one-way sensitivity analysis of device cost showed that ENDURALIFE‑powered CRT‑Ds remained cost saving.
5.10 The company's base case showed that ENDURALIFE‑powered CRT‑Ds cost £22,322 per patient over a 6-year period compared with £27,309 and £29,158 per patient for St Jude Medical and Medtronic CRT‑Ds respectively. The company therefore estimated that using ENDURALIFE‑powered CRT‑Ds would save between £4,987 and £6,836 per patient over 6 years. Cost savings come mainly from fewer replacement procedures.
5.11 The EAC re-ran the company's base case and univariate sensitivity analyses and conducted additional analyses using its preferred estimates. The EAC also did a threshold analysis using the average selling price for ENDURALIFE‑powered CRT‑Ds and allowing the cost of the comparator devices to fall to the point at which each becomes cost neutral. The main changes to the company's model were:
Changes to the list prices of ENDURALIFE‑powered CRT‑Ds and both comparators.
Using warranty data from the comparator manufacturers instead of that from Boston Scientific.
Using NHS reference costs instead of Payment-by-Results tariff costs.
Changes to the sensitivity analysis for complication rates (infection), based on the results of a large Danish cohort study (Kirkfeldt et al. 2014). This changed the infection rate from 2.4% to 0.6% for new implants.
Using patient survival data from the National Institute for Cardiovascular Outcomes Research (NICOR) instead of from Yao et al. (2007).
5.12 The results of the EAC analysis suggested that changing the device cost in the model to the lowest and highest list price for each of the 3 manufacturers results in ENDURALIFE‑powered CRT‑Ds becoming more costly than those from Medtronic, but remaining cost saving compared with those from St Jude Medical.
5.13 The threshold analysis investigated the effect of introducing a price difference between the devices, and calculated the threshold at which ENDURALIFE‑powered CRT‑Ds become cost incurring compared with the comparators. The results showed that, using the same cost of implanting and replacing the CRT‑D as used in the company's base case, ENDURALIFE‑powered CRT‑Ds become cost incurring when they are £4,858 more expensive to purchase than Medtronic CRT‑Ds and £3,858 more expensive to purchase than St Jude Medical CRT‑Ds, with all other model inputs unchanged.
5.14 Using NHS reference costs instead of the Payment-by-Results tariff increased the cost of ENDURALIFE‑powered CRT‑Ds from £22,322 in the company's base case to £30,957. ENDURALIFE‑powered CRT‑Ds remain cost saving compared with the comparators but to a lesser extent than in the company's base case.
5.15 Substituting the actual warranty information supplied by the manufacturers into the model showed that ENDURALIFE‑powered CRT‑Ds remained cost saving.
5.16 Changing the rate of infection for new implants from 2.4% to 0.6% had little effect on the costs.
5.17 Following expert advice, the EAC contacted NICOR, which holds a registry of NHS patients who have had CRT‑Ds implanted, including data on overall survival.
5.18 Using patient survival data from NICOR to replace that from Yao et al. (2007), ENDURALIFE‑powered CRT‑Ds remained cost saving when using the company's base-case device cost. At the lowest and highest list prices, ENDURALIFE‑powered CRT‑Ds become more costly than those from Medtronic, but remain cost saving compared with those from St Jude Medical.
5.19 The EAC concluded that the main driver of the cost model was device price.
5.20 The committee considered that the 6-year time horizon made the cost case uncertain. The EAC was therefore asked to carry out further analyses extrapolating the data available over a patient's lifetime (sections 5.20 to 5.24).
5.21 NICOR provided unpublished data in confidence which showed patient survival by age group after primary implantation of a CRT‑D. The EAC extrapolated CRT‑D lifespan to 15 years using a survival profile for comparator devices: this took an average distribution based on Medtronic and St Jude Medical CRT‑D lifespans reported in Landolina et al. (2015), and then applied the average distribution to the ENDURALIFE‑powered CRT‑Ds from the point at which the ENDURALIFE‑powered CRT‑Ds begin to reach the elective replacement indicator (ERI), at 5 years following implantation.
5.22 The EAC extrapolated patient survival to 15 years using NICOR data for patients aged 50 to 84 years at primary implantation.
5.23 Using the average selling price in the company's base case and the extrapolated data outlined above, the results showed that ENDURALIFE‑powered CRT‑Ds cost £28,234 per patient over 15 years compared with £30,354 and £33,861 per patient for St Jude Medical and Medtronic CRT‑Ds respectively. The EAC therefore estimated that using ENDURALIFE‑powered CRT‑Ds could save between £2,120 and £5,627 per patient over 15 years.
5.24 A threshold analysis investigated the effect of allowing a price difference between the devices, and calculated the threshold at which ENDURALIFE‑powered CRT‑Ds become cost incurring compared with the comparators. The results showed that, using the same cost of implanting and replacing the CRT‑D as used in the company's base case, ENDURALIFE‑powered CRT‑Ds become cost incurring when they are £3,304 more expensive to purchase than Medtronic CRT‑Ds and £1,404 more expensive to purchase than St Jude Medical CRT‑Ds.
5.25 The EAC considered that the 6-year time horizon used in the model may overestimate the potential cost saving of a slightly longer-lasting device. It concluded that a time horizon over the patient's lifetime may be more appropriate.
5.26 The committee was advised that device costs were accurately reflected in the company's base case, which used average selling prices, and that prices are similar between manufacturers. List prices are not a true reflection of what the NHS pays for CRT‑Ds.
5.27 The committee concluded that it would be difficult to ascertain actual NHS device costs for ENDURALIFE‑powered and comparator CRT‑Ds. The EAC was asked to carry out a differential cost threshold analysis to overcome some of these uncertainties.
5.28 The committee accepted that the EAC's revisions to the company's cost modelling provided the most plausible estimates for the cost consequences of adopting ENDURALIFE‑powered CRT‑Ds.