5.1 The sponsor submitted a de novo cost analysis for Pipeline for the treatment of complex intracranial aneurysms. Full details of all cost evidence and modelling considered by the committee are available in the assessment report overview.
5.2 The sponsor presented a decision tree (for the peri-procedural period) followed by a Markov model (for long-term outcomes) to estimate the costs and consequences associated with Pipeline against 5 comparator interventions: stent-assisted coiling, neurosurgical clipping, endovascular parent vessel occlusion, neurosurgical parent vessel occlusion and conservative management. The patient population included those with unruptured large or giant intracranial aneurysms as outlined in the scope, but did not include fusiform or wide-necked aneurysms.
5.3 The decision-tree structure separated patients who had survived initial treatment, based on a mortality rate for the procedure, into 1 of 3 occlusion categories (complete occlusion, residual neck and residual aneurysm). For each occlusion category, patients were tracked through 5 possible health states: 'no complications', 'new non-fatal rupture', 'post rupture', 'fatal rupture' and 'dead (all cause)'. It was assumed that transition probabilities for the health states would be constant over time. The time horizon of the base-case analysis was 10 years.
5.4 An NHS and personal social services perspective was used. The cost analysis included the costs associated with the duration of the procedure, staff time (surgeon, radiologist, nurse, anaesthetist), hospital costs (neurology operating or neurosurgical operating room, and recovery ward), imaging (angiography, fluoroscopy or MRI), consumables, drugs; and for conservative management only, long-term monitoring with annual MRI. The cost associated with stroke was assumed to be representative of the cost of rupture. Costs applied to each type of retreatment were assumed to be the same as the full cost if that treatment had been used initially.
5.5 The costs and consequences associated with adverse events were not included in the base-case analysis because the sponsor considered there to be insufficient reliable data for each treatment group. However, the sponsor did include costs associated with mortality at 31 days, rupture and retreatment.
5.6 In the base case the number of Pipeline embolisation devices used was 1.46, based on data submitted to the sponsor from use of the device in UK hospitals up to August 2011. The number of coils used in the base case was 40 and was derived from an estimate in an editorial review by Wehman in 2006. It was assumed that 1 stent would be needed for each stent-assisted coiling intervention.
5.7 The base case presented the total procedure costs over the 10-year time horizon associated with Pipeline as £24,341. For the comparators, the total procedure costs were £37,451 for stent-assisted coiling, £11,658 for neurosurgical clipping, £16,893 for endovascular parent vessel occlusion, £11,654 for neurosurgical parent vessel occlusion and £10,352 for conservative management. The only intervention against which Pipeline was shown to be cost saving was stent-assisted coiling, with a cost saving of £13,110 per patient.
5.8 Two scenario analyses were presented. One included costs associated with the adverse events of subarachnoid haemorrhage related to the aneurysm, thromboembolic stroke and intracranial haemorrhagic stroke remote from the aneurysm using data from the PUFS study and data from Darsault et al. (2001) for the comparators. The other scenario analysis restricted the time horizon to 6 months (short-term). Conservative management was excluded from the short-term scenario because it does not have a 'peri-procedural' mortality rate.
5.9 In both scenario analyses Pipeline was shown to be cost saving only when compared with stent-assisted coiling. When costs associated with adverse events were included in the model, Pipeline remained a cost-saving intervention compared with stent-assisted coiling, with an associated saving of £13,327. When outcomes were restricted to the short term (6 months), Pipeline remained cost saving compared with stent-assisted coiling (£10,316).
5.10 Sensitivity analyses carried out by the sponsor showed that the main factors influencing the cost analysis were the number and cost of consumables, in particular the numbers of Pipeline embolisation devices and endovascular coils. The sponsor carried out sensitivity analysis for the use of 1–3 Pipeline embolisation devices and separately for 5–100 coils.
5.11 The number of Pipeline embolisation devices used in the base case was 1.46 per patient. On receipt of more data from UK hospitals, the sponsor submitted a revised number of 1.658 in October 2011. The external assessment centre reviewed the data and concluded that when 1.658 Pipeline embolisation devices were used, Pipeline was more costly compared with stent-assisted coiling if 22 coils were used (an estimated cost increase of £19), but cost saving when 23 coils were used. The cost saving when using 1.658 Pipeline embolisation devices compared with stent-assisted coiling with 23 coils was estimated to be £588 (total procedure costs of £26,546 and £27,134 respectively). When 2 Pipeline embolisation devices were used, Pipeline was more costly by an estimated £185 than stent-assisted coiling using 28 coils, but was less costly than stent-assisted coiling when 29 coils were used. The cost saving when using 2 Pipeline embolisation devices compared with stent-assisted coiling with 29 coils was estimated to be £421 (total procedure costs of £30,354 and £30,775 respectively).
5.12 During consultation, expert advisers expressed doubts about 5 parameters in the cost model: the cost of microcatheters; the use of a balloon in stent-assisted coiling; drug costs; the procedure duration; and the cost of additional endovascular equipment. Additional exploratory analyses were carried out by the external assessment centre to examine the impact of changing these parameters, in ways which the expert advisers suggested were more appropriate.
5.13 In the sponsor's economic model, the cost of microcatheters used in stent-assisted coiling was comparable to the cost of a Marksman catheter (£1,030) used to insert Pipeline. However, the expert advisers stated that cheaper microcatheters (average cost £460.50) would be used for stent-assisted coiling than for Pipeline in UK practice. The external assessment centre's exploratory analyses demonstrated that reducing the cost of 2 microcatheters for stent-assisted coiling from £1,030 to £460.50 reduced the total procedure cost for stent-assisted coiling from £37,451 to £36,137.
5.14 The sponsor's economic modelling also assumed that a balloon is used in 50% of stent-assisted coiling procedures and is not used to insert Pipeline. However, the expert advisers stated that use of a balloon is relatively uncommon in the UK for stent-assisted coiling procedures. In the exploratory analyses carried out by the external assessment centre, removing the cost of a balloon reduced the total procedure cost for stent-assisted coiling from £36,137 to £35,725, while the total procedure cost for Pipeline (with 2 devices) remained at £30,354.
5.15 The sponsor's economic model assumed different drug use for patients having stent-assisted coiling and those having Pipeline (18,000 mg aspirin and 6,750 mg clopidogrel compared with 25,000 mg aspirin and 13,500 mg clopidogrel respectively). The expert advisers stated that in clinical practice drug use in the 2 groups is likely to be equal. The external assessment centre noted that in the sponsor's economic model drug use was calculated from 2 non-comparative studies, but that few published data are available on clinical use. The external assessment centre stated that the low cost of these drugs means that relatively small differences in drug use are unlikely to have a significant impact on the overall procedure costs for stent-assisted coiling and Pipeline. In the absence of a systematic review the external assessment centre stated that the drug costs used in the model were appropriate.
5.16 In its exploratory analysis the external assessment centre identified 2 calculation errors in the sponsor's economic model, both relating to the number of days of drug therapy. The external assessment centre stated that correcting these calculation errors made only a very small difference in the total procedure costs for both Pipeline and stent-assisted coiling, reducing them from £30,354 to £30,346 and £35,725 to £35,724 respectively.
5.17 In the sponsor's economic model, the average duration of the procedure (in hours) and therefore the use of additional endovascular equipment were different for stent-assisted coiling and insertion of Pipeline. The expert advisers stated that the procedures are likely to take the same amount of time in clinical practice. The external assessment centre noted that the time taken for stent-assisted coiling was derived from 1 non-comparative study and the time for inserting Pipeline from another. However the external assessment centre was not able to judge which study was the best source for duration of procedure and no systematic review was carried out. The external assessment centre stated that because both studies were appropriately and accurately used in the model no changes were justified with regard to procedure time for stent-assisted coiling or Pipeline. The external assessment centre also noted that the sponsor included the use of additional endovascular equipment in the model based on the duration of procedure, but provided no justification for why it was included only for stent-assisted coiling and not for Pipeline or the other comparators. Furthermore, there was no justification for why the cost of additional equipment was only included in calculating the retreatment cost but not the cost of the initial procedure. The exploratory analyses demonstrated that removing the costs of additional endovascular equipment from the model reduced the total procedure cost of stent-assisted coiling from £35,724 to £35,693 while the total procedure cost for Pipeline remained at £30,346.
5.18 The committee considered the cost consequences of using Pipeline against 5 comparators – stent-assisted coiling, neurosurgical clipping, endovascular parent vessel occlusion, neurosurgical parent vessel occlusion and conservative management.
5.19 The committee noted that in UK clinical practice, patients who might currently be considered for Pipeline would be those for whom surgery would not be possible and for whom stent-assisted coiling would be the only other potential intervention. It therefore considered that comparison of costs with those for stent-assisted coiling was of particular relevance. The committee noted that for this cost comparison, the main drivers of cost were the numbers of Pipeline embolisation devices used and the numbers of coils used during stent-assisted coiling. The committee was given advice from the experts about whether 1 or 2 Pipeline embolisation devices should be used as the most typical number on which to base its judgements about cost. It therefore considered the cost modelling for a range of scenarios – with different numbers of Pipeline embolisation devices. During its discussions the committee was informed both by expert advisers and by data from the sponsor that 2 Pipeline embolisation devices were likely to be needed for most patients treated in the UK for complex giant or large aneurysms.
5.20 There was considerable uncertainty about the number of coils likely to be used during stent-assisted coiling. The committee was advised that 40 coils was probably an overestimate of the number needed for most complex aneurysms and there were suggestions that about 25–30 coils might be more typical of the number needed for aneurysms being considered for treatment using Pipeline. The committee recognised that it is not possible to estimate accurately the number of coils which will be required to treat an aneurysm during stent-assisted coiling, but it was advised by experts that the number can usually be predicted with an accuracy of plus or minus 6 coils.
5.21 The committee discussed a number of parameters in the sponsor's economic model based on comments received during public consultation. As described in sections 5.12 to 5.17, these included: the cost of microcatheters used for stent-assisted coiling; balloon use in UK practice during stent-assisted coiling; drug costs for patients having Pipeline and those having stent-assisted coiling; the duration of the procedure and the use of additional endovascular equipment in stent-assisted coiling. These parameters were examined in additional analyses by the external assessment centre.
5.22 The committee considered that the costs for microcatheters used for stent-assisted coiling in the sponsor's base-case analysis were overestimated. It agreed with the views of the expert advisers that it was reasonable to use the average cost of the microcatheters most commonly used in UK clinical practice. This reduced the total procedure cost for stent-assisted coiling from £37,451 to £36,137.
5.23 Based on expert advice, the committee accepted that the inclusion of balloon use during stent-assisted coiling in the sponsor's base-case analysis was inappropriate and that it was reasonable to adjust this cost to zero. This reduced the total procedure cost for stent-assisted coiling from £36,137 to £35,725, while the total procedure cost for Pipeline remained at £30,354.
5.24 The committee considered the drug costs presented in the sponsor's base-case analysis. It noted the external assessment centre's view that few published data are available on clinical use and that the low cost of these drugs and the difference in drug use would not be likely to have any significant impact on the overall procedure costs for either stent-assisted coiling or Pipeline. The committee agreed that the drug cost in the sponsor's base-case analysis was appropriate
5.25 The committee considered the average duration of the procedure presented in the sponsor's base-case analysis. It noted the external assessment centre's view that both studies were appropriately and accurately used in the model and that therefore no changes were justified for the average duration of the procedure for stent-assisted coiling or inserting Pipeline. The committee agreed that the average duration of the procedure presented in the sponsor's base-case analysis was appropriate.
5.26 The committee was advised by the experts that it was not appropriate to include the cost for additional endovascular equipment for stent-assisted coiling only. This was because the cost of endovascular equipment was likely to be the same for Pipeline compared with stent-assisted coiling. The committee accepted the views of the expert advisers that removing this cost from the model for stent-assisted coiling was appropriate. This reduced the total procedure cost for stent-assisted coiling from £35,724 (this cost takes account of the drug cost calculation errors referred to in section 5.16) to £35,693 while the total procedure cost for Pipeline remained at £30,346.
5.27 With these issues in mind, the committee considered a graph showing the costs of Pipeline and of stent-assisted coiling using different numbers of Pipeline embolisation devices and coils. This indicated that the total cost of treatment using 2 Pipeline embolisation devices is greater than that of stent-assisted coiling (using 1 stent) if 31 coils or fewer are used, but Pipeline is less costly if 32 or more coils are needed. The cost saving associated with Pipeline compared with stent-assisted coiling with 32 coils was estimated to be £492 (total costs of £30,346 and £30,838 respectively).
5.28 The committee noted that using 2 Pipeline embolisation devices would incur a higher total treatment cost than neurosurgical clipping, endovascular parent vessel occlusion, neurosurgical parent vessel occlusion or conservative management for patients in whom those other options were feasible.
5.29 For the guidance review, the external assessment centre revised the model to reflect updated costs. The main parameter changes were costs associated with staff, hospital imaging equipment, drugs, rupture and adverse event costs. Further details of the revised model are in the revised model summary.