Background
Cluster headache (CH) is a debilitating condition associated with intense pain and cranial autonomic symptoms, which cause marked disability [
1]. The disorder adversely affects quality of life [
2] and is associated with substantial health care costs (more than €11,000 per year) [
3]. The condition can be chronic or episodic. Both direct costs (e.g. medication, clinic visits) and indirect costs (e.g. reduced work capacity) have been found to be substantially higher for patients with chronic CH (cCH) than for those with episodic CH [
3]. Few drugs (e.g. subcutaneous [SC] sumatriptan, intranasal [IN] zolmitriptan, and dihydroergotamine [DHE] injection) are approved by various regulatory agencies for abortive treatment [
4,
5]. Lithium is approved for CH prophylaxis in Germany [
6] and is used off-label in other areas. Other agents such as verapamil and topiramate are also used off-label despite a lack of rigorous, well-controlled studies to support their use in the prevention of CH attacks [
7‐
9]. Although short-term methylprednisone therapy may be effective in CH prophylaxis, several safety concerns preclude its long-term use [
8].
Vagus nerve stimulation (VNS) is a neuromodulatory technique that is well established for epilepsy and depression and has been applied to a variety of other disorders including Alzheimer disease, migraine, and CH [
10‐
12]. It is thought to suppress pain through inhibition of vagal afferents in the trigeminal nucleus caudalis (TNC) [
13] and by blocking or reversing increases in TNC glutamate levels [
14]; VNS has also been implicated in modulation of the cholinergic anti-inflammatory pathway [
15‐
17].
In an initial open-label study (
N = 19), non-invasive vagus nerve stimulation (nVNS) was found to be effective in the prevention and treatment of CH [
11]. Subsequently, a larger (
N = 97), prospective, open-label, randomised study (PREVA [
18]) evaluated the safety and efficacy of adjunctive treatment with a novel nVNS device (gammaCore
®) in patients with cCH. In the PREVA trial, compared with standard of care (SoC) alone, adjunctive nVNS (nVNS + SoC) was associated with significantly greater decreases from baseline in the number of CH attacks per week and the use of abortive medications. Compared with SoC alone, nVNS + SoC was also associated with a significantly higher response rate (i.e. the proportion of participants with a ≥50 % reduction from baseline in the number of CH attacks per week; 40 % for nVNS + SoC vs 8.3 % for SoC alone,
P < 0.001) and significantly greater improvements from baseline in quality-of-life measures, with no serious treatment-related adverse events.
The present analysis was undertaken to quantify the economic impact of nVNS therapy in patients with cCH. By developing a pharmacoeconomic model and applying it to data from the PREVA study, we evaluated whether nVNS is a cost-effective treatment option compared with the current standard practice in a European setting. Analysis using German costs is the focus of this paper because Germans represented the largest proportion of PREVA participants. To corroborate our findings and widen their applicability, we conducted a similar analysis using UK costs, which is briefly described in the
Discussion section.
Discussion
The treatment of CH is challenging, and many of the commonly used abortive and preventive medications are associated with serious safety risks, poor tolerability, and/or marginal efficacy. For acute treatment, triptans are contraindicated in patients with cardiovascular disease [
25,
26]. Drug costs or restrictions on prescribing and/or coverage may further limit triptan accessibility for many patients [
27,
28]. Long-term frequent use of triptans, as may be needed for cCH management, can in turn lead to the development of medication overuse headache [
29,
30], which, although rare, has been reported in patients with CH [
31,
32]. Oxygen may delay rather than abort CH attacks in some patients and has portability limitations [
25,
26], and DHE may be associated with fibrosis (e.g. cardiac, pulmonary, pleural), ergotism, and chest tightness [
26,
33]. For prophylactic treatment, verapamil has a high potential for drug interactions, and the large dosages required for CH treatment are associated with adverse cardiac events such as arrhythmias, as well as oedema [
26]. Lithium requires progressive titration and frequent drug-level monitoring because of its narrow therapeutic window and the risk of toxicity [
25,
26,
34], and topiramate is often poorly tolerated owing to its cognitive side effects [
26]. Thus, more practical and cost-effective treatment approaches for CH are needed. Results from the PREVA study [
18] suggest that in addition to reducing the frequency of CH attacks, adjunctive nVNS therapy may decrease the need for abortive medications and improve quality of life in patients with cCH. The current pharmacoeconomic analysis indicates that adjunctive nVNS is likely to result in cost savings when compared with SoC alone. Notably, the present analysis was conservative in that it included only the costs associated with use of abortive medications without accounting for other potential sources of cost savings (e.g. reduced frequency of clinic visits, fewer hospitalisations, increased productivity).
Currently, there are few good options for acute or prophylactic treatment of CH. Neuromodulation methods such as sphenopalatine ganglion (SPG) stimulation and occipital nerve stimulation (ONS) have shown some promise in CH prevention, but most studies of these techniques have been small and/or have lacked control arms [
35,
36]. Furthermore, SPG and ONS are invasive, expensive, and associated with risks inherent with implanted devices (e.g. infection, pain at the site of implantation, electrode migration). The findings that nVNS is effective in cCH prophylaxis [
18], is not associated with risks that are inherent in invasive neuromodulation methods, and offers cost savings over the current standard practice suggest that this therapy warrants a prominent place in the management of cCH.
The current analysis is subject to certain limitations. The PREVA study provided data from an 8-week period, which were extrapolated to assess cost-effectiveness over 1 year. Although there have been few cost-effectiveness evaluations of neuromodulatory techniques for the treatment of primary headache disorders, such studies have generally included time horizons of at least 3 years [
37‐
39]. Considering the time frame of PREVA, a 1-year time horizon was chosen for this analysis to preserve robustness and to avoid introducing unnecessary uncertainty. As in patients with epilepsy [
40], evidence suggests that patients with headache may have improved response to VNS with longer-term treatment [
41,
42]. Although increases in response rate with long-term VNS have yet to be explored in CH, the current analysis could be viewed as conservative because the duration of PREVA may not have allowed demonstration of the full benefit of nVNS.
Recently, the National Institute for Health and Care Excellence (NICE) Interventional Procedures Advisory Committee noted that the relapsing/remitting nature of CH and migraines as well as the potential for placebo effects should be considered when interpreting evidence of treatment efficacy for these conditions [
43]. Indeed, because periods of relapse and remission are common among patients with primary headache disorders, research in this area may be susceptible to regression artefacts [
44,
45]. However, the PREVA study included data from patients with cCH only. By
International Classification of Headache Disorders definition [
46], cCH is not associated with extended periods of remission (i.e. ≥1 month), suggesting that the phenomenon of regression to the mean (e.g. aberrantly high attack frequency at baseline followed by a decrease in attack frequency regardless of treatment group) would not be expected. Because the PREVA study lacked a sham treatment group, the degree to which the placebo effect might have contributed to the cost-effectiveness of nVNS is unclear. Nevertheless, the clinically relevant design of the PREVA study was valuable in that it allowed for observation of medication use in a control group that likely reflects real-world use.
As with any probabilistic analysis, some degree of uncertainty is inherent in the current investigation. To address this, a sensitivity analysis and a range of alternative scenarios were included, and results from all of these suggested that nVNS + SoC was more effective and cost saving than SoC alone. Results were relatively insensitive to assumptions about late responders in the nVNS + SoC arm. In the sensitivity analysis, where the 4 late-responding patients were classified as non-responders, nVNS + SoC was dominant over SoC alone in all modelled scenarios.
The current analysis cannot be directly extrapolated across all of Europe because it evaluates cost-effectiveness from a German health insurance perspective. To explore the generalisability of our findings, we conducted the same analysis from a UK perspective and found similar results. For the base case, the probabilistic analysis resulted in mean expected costs of £5409.83 for nVNS + SoC and £5393.31 for SoC alone and mean QALY of 0.538 for nVNS + SoC and 0.438 for SoC alone. The incremental cost-effectiveness ratio of nVNS + SoC was £166.12, and 47 % of the probabilistic simulations resulted in cost savings for nVNS + SoC over SoC alone (J. Morris, unpublished data, 2016). The degree to which these results can be generalised to other countries may vary depending on specific drug prices and the availability of generic medications in those markets.
Lastly, the current cost-effectiveness projections included only the costs associated with the use of abortive treatments. This suggests that our analysis is conservative, as data on additional health care resource use (e.g. clinic visits) would likely lead to a disproportionate cost increase for the SoC-alone group. Likewise, potential health benefits from decreased use of abortive medications (e.g. drug-related side effects) and effects on indirect costs (e.g. increased work capacity), which could further enhance the economic profile of nVNS, were not considered herein. The economic benefits of nVNS could be established with greater certainty by incorporating additional cost components into future studies.
Competing interests
James Morris is an employee of Cogentia Healthcare Consulting Ltd.
Andreas Straube has received honoraria for educational talks and advisory boards from Allergan Germany; Boehringer Ingelheim; Cerbotech; Desitin Pharma; electroCore, LLC; Hormosan Pharma; MSD Germany; and Teva.
Hans-Christoph Diener has received honoraria for participation in clinical trials, contribution to advisory boards, or oral presentations from Addex Pharmaceuticals; Alder Biopharmaceuticals; Allergan, Inc.; Almirall, SA; Amgen, Inc.; AstraZeneca; Autonomic Technologies, Inc.; Bayer Vital; Berlin-Chemie; Boehringer Ingelheim; Bristol-Myers Squibb; Chordate Medical; Coherex Medical; CoLucid Pharmaceuticals, Inc.; electroCore, LLC; GlaxoSmithKline; Grünenthal; Janssen-Cilag; Johnson & Johnson; Labrys Biologics Inc.; La Roche; Lilly; 3 M; Medtronic; Menarini Pharma; Minster Pharmaceuticals; MSD; NeuroScore; Novartis; Pfizer, Inc.; Pharma Medica Research Inc.; Pierre Fabre Laboratories; Sanofi; Schaper & Brümmer; St. Jude Medical; Teva; and Weber & Weber. Financial support for research projects has been provided by Allergan, Inc.; Almirall; AstraZeneca; Bayer; electroCore, LLC; GlaxoSmithKline; Janssen-Cilag; MSD; and Pfizer, Inc. Headache research at the Department of Neurology in Essen is supported by the German Research Council (DFG), the German Ministry of Education and Research (BMBF), and the European Union. Dr. Diener has no ownership interest in and does not own stocks of any pharmaceutical company.
Fayyaz Ahmed has nothing to disclose.
Nicholas Silver has received honoraria from Allergan, Inc., and electroCore, LLC; investigator fees from Amgen Inc. and Eli Lilly and Company; and investigator fees paid to the Walton Centre.
Simon Walker is an employee of Cogentia Healthcare Consulting Ltd.
Eric Liebler is an employee of electroCore, LLC, and receives stock ownership.
Charly Gaul has received honoraria from Allergan, Inc.; Autonomic Technologies, Inc.; Bayer; Berlin-Chemie; Boehringer Ingelheim; Desitin Pharmaceuticals; electroCore, LLC; Grünenthal; Hormosan Pharma; MSD; and St. Jude Medical. Dr. Gaul has no ownership interests and does not own any pharmaceutical company stocks.
Authors’ contributions
JM contributed to the design and construction of the pharmacoeconomic model and data analysis and interpretation, as well as to drafting and revision of the manuscript. AS, H-CD, FA, and NS were principal investigators in the PREVA clinical study and provided clinical expertise in data interpretation and revision of the manuscript. SW contributed to the design and construction of the pharmacoeconomic model, data analysis and interpretation, and revision of the manuscript. EL contributed to the design and construction of the pharmacoeconomic model and data interpretation, as well as to drafting and revision of the manuscript. CG was the primary principal investigator in the PREVA clinical study and provided clinical expertise in data interpretation and revision of the manuscript. All authors read and approved the final manuscript draft.