3.1 The evidence for Senza considered by the external assessment centre (EAC) comprised 10 studies in adults with chronic neuropathic pain. Of these 10, 3 studies (Al-Kaisy et al. 2017b, De Andres et al. 2017 and Van Buyten et al. 2017) became available during consultation on the original draft recommendations, and 1 became available (Amirdelfan et al. 2018) after a further consultation on the second draft recommendations. The 10 studies were:
2 randomised controlled trials comparing Senza and low‑frequency spinal cord stimulation (SCS; Kapural et al. 2015 and 2016 and De Andres et al. 2017)
1 before-and-after study (Tiede et al. 2013)
5 single-arm observational studies (Al-Kaisy et al. 2014, Russo et al. 2016, Rapcan et al. 2015, Al-Kaisy et al. 2017a and Al-Kaisy et al. 2017b)
1 retrospective chart review (Van Buyten et al. 2017)
1 quality-of-life analysis using data from Kapural et al. 2016 (Amirdelfan et al. 2018).
For full details of the clinical evidence, see section 3 of the assessment report and the supplementary EAC documents.
3.2 The EAC initially considered Kapural et al. (2016) to be the most relevant study providing the best quality evidence. Although it identified that the study had the potential for performance, detection and reporting bias, the EAC was satisfied that the trial's limitations did not affect the overall direction of the results.
3.3 The other 5 studies initially identified were single-arm observational studies, the results of which generally supported and corroborated the results of Kapural et al. (2016). The highest quality of these was Al-Kaisy et al. (2014), which reported results up to 2 years.
3.4 The EAC initially concluded that the evidence was strong and relevant to the decision problem, and that it showed that Senza provided substantially better pain control than low‑frequency SCS. However, it noted gaps in the evidence base, particularly the lack of long-term studies and the absence of a sham control.
3.5 Following consultation on the draft guidance, 4 additional studies were identified as being relevant to the decision problem: De Andres et al. (2017), Van Buyten et al. (2017), Al-Kaisy et al. (2017b) and Amirdelfan et al. (2018).
3.6 The EAC considered that, in addition to Kapural et al. (2016), the De Andres et al. and Van Buyten et al. studies were most relevant to the decision problem. De Andres et al. reported that Senza and low‑frequency SCS had similar benefits, conflicting with the results of Kapural et al. Van Buyten et al. is a retrospective chart review that reported the rates and reasons for removing SCS devices in 4 centres that had done 955 implantations (155 of which were Senza) in 822 patients.
3.7 Al-Kaisy et al. (2017b) reported extended follow-up data to the original study by the same author which had been included in the company submission assessed by the EAC in its original report. The new data reported that early improvements (up to 12 months) in pain, disability and quality of life were maintained until 36 months.
3.8 The Amirdelfan et al. (2018) study provided data on additional outcomes from Kapural et al. at 12 months. The EAC concluded that this study provided additional evidence that Senza may result in improved patient-reported outcome measures compared with low‑frequency SCS, but did not detect a difference in generic health-related quality-of-life outcomes.
3.9 Kapural et al. (2016) and De Andres et al. (2017) reported randomised controlled trials comparing Senza with low‑frequency SCS, with inconsistent findings. Specifically, Kapural et al. (2016) reported a statistically significant reduction in back and leg pain with Senza compared with low‑frequency SCS whereas De Andres et al. (2017) reported no difference in pain scores. The trial design and conduct of the De Andres study was openly challenged and the authors' responses to these criticisms were summarised in a letter that was published in the same journal as the original paper (see the supplementary EAC documents for more details).
3.10 Having reviewed all of the evidence, the EAC concluded that Senza is likely to be at least as effective as low‑frequency SCS in terms of reducing pain in appropriately selected patients. However, it noted that both the Kapural et al. (2016) and particularly the De Andres et al. (2017) studies were subject to bias and had design and reporting weaknesses, significantly more so for the latter. Because of this, the EAC considered that the results should be interpreted with caution.
3.11 Because of the inconsistent trial results and because of a large number of conflicting comments received during both consultations, a second EAC reviewed the randomised controlled trial evidence. It concluded that the De Andres et al. (2017) study was methodologically worse than Kapural et al. (2016). It drew specific attention to weaknesses in terms of the trial's analysis, governance and design (see the supplementary EAC documents or more details).
3.12 The company's cost model was based on a published cost-effectiveness study (Annemans et al. 2014) comparing Senza separately with conventional medical management, reoperation and low‑frequency SCS devices (both rechargeable and non-rechargeable). The model, which was also used to inform the NICE technology appraisal guidance on spinal cord stimulation (Simpson et al. 2008), was a 2-stage decision analytic model that used a decision tree for the first 6 months, followed by a Markov state transition model with a 15‑year time horizon. For full details of the economic evidence, see section 4 of the assessment report and the supplementary EAC documents.
3.13 The EAC considered Annemans et al. (2014) to be of high quality and the company's cost model to be of good methodological quality. It was therefore initially satisfied with the reported results and sensitivity analyses. However, the publication of Van Buyten et al. (2017) during the first consultation provided additional real-world data as an alternative estimate for the rate of unanticipated explantation used in the cost model. The EAC did not change the anticipated explants parameters in the model but estimated the unanticipated explantation parameters used in the cost model by extrapolating the data available from Van Buyten et al. for explantations because of inadequate pain relief (see the supplementary EAC documents for more details).
3.14 Many of the costs in the model, including the acquisition costs for Senza and its comparators, were adjusted for inflation from the original values in the Annemans et al. study. The EAC considered it inappropriate to inflate drug prices in this way because they are subject to a wide range of non-inflationary factors. The EAC explored this further with 4 hypothetical scenarios to assess how different drug costs affect the cost consequences of using Senza.
3.15 The main drivers of the cost modelling results were acquisition costs, explantation rates and device lifespan, particularly for non-rechargeable SCS devices, which need to be replaced around every 4 years. The company's base-case results showed that, over 15 years, Senza could lead to cost savings of £4,795 compared with rechargeable low‑frequency SCS devices and £7,755 compared with non-rechargeable low‑frequency SCS devices.