Background
Migraine is a chronic, neurologic headache disorder characterised by recurrent attacks of varying intensity, duration and symptoms. In addition to headache and pain, symptoms frequently include deterioration of physical functions, nausea, vomiting, and sensitivity to light and sounds. A migraine attack typically lasts from 4 to 72 h if left untreated [
1]. In the adult Swedish population 17% of women and 10% of men suffer from migraine [
2]. Similarly in Norway, the corresponding is approximately 15% of women and 7% of men [
3].
Migraine is an incapacitating disease with a substantial quality of life (QoL) loss and economic burden. The latter partially through increased healthcare costs, but mainly due to lost productivity and reduced functioning. Migraine-related productivity loss consists of both absenteeism and decreased productivity while on the job (presenteeism). Previous studies have shown a clear relationship between the number of headache days and increased costs and decreased QoL [
4‐
10]. For example, a recent Italian study showed that the direct annual cost of CM patients is 4.8 times higher than the cost of EM patients [
11]. Despite the high prevalence and the apparent and substantial burden, few current health economic studies have explored the real-world impact of migraine and headaches in the Nordics. The Eurolight study – conducted over a decade ago – estimated the cost of headaches in the European Union (EU) to EUR 173 billion annually. Migraine accounted for 64% of the cost, with indirect costs comprising 90% of the total [
12]. A Swedish survey study found that increased monthly migraine days are associated with significant QoL losses and cost increases, with productivity losses comprising the vast majority (80%) of the total [
13]. The study identified potential for significant QoL improvement and cost decreases related to the reduction of migraine days.
Pharmaceutical treatments for migraine in Sweden and Norway include acute medications for symptom relief during migraine attacks and prophylactic medications intended to prevent and reduce severity of future attacks. Acute medications include analgesics, as well as triptans, antiemetics and ergotamine [
1,
14,
15].
Prophylactic treatment is considered for patients with frequent or severe migraine attacks, despite acute treatment. For first choice of prophylactic treatment, beta-blockers are recommended, while other alternatives include antiepileptics, antidepressants, angiotensin 2 blockers or ACE inhibitors, calcium antagonists, angiotensin-II receptor inhibitors or OnabotulinumtoxinA (Botox).
Botox is indicated in both Sweden and Norway for symptom relief in adults fulfilling criteria for chronic migraine (CM) in patients who have responded inadequately or are intolerant of prophylactic migraine medications [
16]. CM is defined as > 3 months of headaches occurring on ≥15 days per month, of which ≥8 days are with migraine [
17]. In Sweden, Botox is reimbursed according to its approved migraine indication. In Norway, Botox is reimbursed for CM in patients who previously tried a beta-blocker (propranolol, metoprolol or atenolol) and topiramate.
Calcitonin gene-related peptide (CGRP) inhibitors erenumab (Aimovig), galcanezumab (Emgality) and fremanezumab (Ajovy) are indicated for prevention of migraine in adults with at least four migraine days per month in both Sweden and Norway. In Sweden, erenumab and fremanezumab are reimbursed in patients with CM who have responded insufficiently or are intolerant to at least two prophylactic migraine treatments. Galcanezumab is not available in Sweden. In Norway, erenumab, fremanezumab and galcanezumab are reimbursed for CM in patients who have responded insufficiently or are intolerant to at least three different classes of prophylactic migraine treatments.
Several clinical and real-world studies have shown that treatment with Botox reduces the number of headache days in patients with CM as well as days absent from work and school [
18‐
21]. The safety and efficacy of Botox in CM in adult patients was evaluated in randomized, double-blind, placebo-controlled clinical trials PREEMPT 1 and 2: Phase 3 REsearch Evaluating Migraine Prophylaxis Therapy (PREEMPT) – the largest clinical program investigating the use of Botox as prophylactic treatment for CM [
18]. International, multicentre, open-label long-term prospective study COMPEL (The Chronic Migraine OnabotulinumtoxinA Prolonged Efficacy open Label) provided additional clinical evidence of efficacy and long-term safety and tolerability [
22]. European, open-label, multicentre, prospective, noninterventional study REPOSE provided evidence of a sustained reduction in headache-day frequency and significant improvements in QoL in CM patients treated with Botox [
23].
While previous cost-effectiveness (CE) analyses of Botox in chronic migraine have been conducted in the United Kingdom (UK) [
24,
25] and Italy [
26] there is a lack of Nordic CE studies. CE outcomes are highly dependent on country specific factors such as e.g. national health system, clinical practice and local cost and utility data. Therefore, previous European CE studies are inadequate for drawing conclusions on the Nordic situation. This study described the economic consequences of migraine in Sweden using cost of illness (COI) survey data. Additionally, results from the COI study were used to populate a CE analysis to assess the CE of Botox for the treatment of CM in Sweden and Norway.
Discussion
Our COI findings show a clear correlation between headache frequency and increased costs, indicating substantial potential cost savings related to reducing the number of headache days per month. Botox reduces the number of headache days per month in a patient group with few other treatments available. And, as made apparent by the cost-effectiveness analysis, Botox is likely a cost-effective treatment option in Sweden and Norway.
The correlation between productivity losses – both in terms of labour productivity and foregone household work – and headache frequency can be interpreted as a proxy for quality of life. The EQ-VAS results further support this notion and demonstrate that the number of headache days per month is not only a significant cost driver, but also induces a substantial quality of life burden for the sufferers. EQ-VAS in the 25–31 headache days group was almost halved in comparison to the general population, while EQ-VAS in the 0–4 headache days group was almost at general population level. A large majority of the cost burden of migraine was due to indirect costs resulting from reduced productivity and functional ability (ranging from 82% of total costs in the 0–4 headache days group to 91% in the 25–31 headache days group). These results are in line with previous findings [
12,
13,
57]. The CE findings indicate that Botox treatment is cost effective in Sweden and Norway, with ICERs well below the previously discussed TLV and NOMA WTP thresholds (ranging from approximately EUR 26,000 to EUR 34,000 in Sweden and from EUR 35,000 to 46,000 in Norway). The results were robust to several sensitivity analyses but were the most sensitive to the inclusion of indirect costs, assumptions on the stopping rule, shortening the time horizon and variations in utilities. The results of this analysis are in line with previous results. A 2013 UK study assessed the CE of Botox in adults in CM compared to placebo, and estimated an ICER of GBP 15,028, assuming a 2-year time horizon and not including indirect costs. When including indirect costs the ICER was reduced to GBP 9422 [
24]. An updated analysis resulted in an ICER of GBP 16,306, in CM patients who have previously failed at least three preventive treatments [
25]. In 2014, an Italian study comparing Botox to placebo in patients with CM estimated an ICER of EUR 9407, assuming a 2-year time horizon and not including indirect costs. When considering indirect costs, the ICER was reduced to EUR 815 [
26]. Likely drivers of differences in these results are utility values used, the country specific Botox price, different assumptions regarding stopping rule, and our relatively higher costs related to productivity losses.
The COI study had several limitations. As inherent with survey studies, there is risk of recall bias. The relatively short time span for recalling headaches and resource use (i.e. 3 months) likely reduced this risk. Additionally, the fact that the 3 months responses were extrapolated to annual costs adds uncertainty as the extrapolation relies on the underlying assumption that migraine severity and intensity is fairly constant over one-year periods. The survey being undertaken in a patient organisation reduces the generalisability of the results, as there is a risk that these patients differ from the general migraine population. Patients may be motivated to participate in such organizations due to large disease burden, resulting in a patient population with heightened severity as compared to general disease populations. Regarding this study, this is likely not a major concern considering the results being stratified over disease severity expressed in monthly average number of headache days. Additionally, several local clinical experts have confirmed that the patient characteristics of our sample corresponds well with the clinical practice migraine patient population.
This study only included patients with diagnosed migraine. Norwegian estimations suggest that approximately 50% of migraine sufferers are undiagnosed [
58]. There is however no straightforward way to include these undetected patients. In order to limit the study to actual migraine sufferers, including only diagnosed migraine was the only viable option. The limited patient sample (
N = 454), and especially when stratified over the headache days subgroups present a statistical challenge made apparent by the relatively high standard deviations obtained for each of the healthcare utilization estimates. Concerns raised regarding the robustness of the results are however reduced by the fact that results are in line with previous literature. A final limitation with the COI study is that results likely underestimated the indirect costs as the survey did not include a question on workforce participation. This renders a risk that patients completely incapacitated by their headache and therefore not in the workforce may have not answered the question related to absenteeism. A proportion of these patients may have answered the questions relating to foregone housework instead, further augmenting the importance of considering unpaid labour and spare time in the analysis. This concern is increased by the fact that our findings regarding productivity losses was lower than previous Swedish results [
59], which was also pointed out by clinical experts. It should therefore be noted that the results regarding indirect costs are at risk of being underestimated, and that the total disease costs may be higher than estimated in our study. Additionally, this study utilised conservative approaches in the costing of both presenteeism and foregone household work.
Uncertainty is inherent in all CE modelling. In order to reduce uncertainty related to assumptions and model inputs extensive interviews with local clinical experts were undertaken. Limitations however remained. The CE model was constructed with headache frequency intervals based on the Botox clinical trials, and therefore differed slightly from the approach used in the COI estimations. This was solved by estimating the model inputs sourced from the COI study using the model intervals. The comparator, placebo, was selected in absence of an established standard of care, and to be in line with the Botox clinical trials. The choice was validated by local clinical experts. The PREEMPT placebo patients received saline injections and were permitted to remain on their standard acute headache treatment. The trials observed considerable headache frequency reductions in the placebo group [
18‐
21] which are not likely to be observed in a real-world setting, making the choice of comparator conservative. No CE analyses comparing Botox to CGRP inhibitors (erenumab and fremanezumab) were undertaken due to lack of reliable comparative efficacy data and due to lack of price information: erenumab and fremanezumab have been launched in the Nordics and are reimbursed in accordance with confidential price agreements. However, as suggested by TLV, an indirect comparison could be done by comparing the ICERs from the analyses comparing each active substance to placebo (given similar QALY differences) [
48]. For Sweden, TLV reports the difference in QALYs gained to be 0.20 and 0.21 for fremanezumab and erenumab, respectively, vs placebo, resulting in ICERs of SEK 505,000 (EUR 47,690) and SEK 557,000 (EUR 52,601), respectively [
48]. Both analyses used confidential net prices. For Norway, NoMA reports the difference in QALYs to be 0.17 for erenumab vs placebo (ICER not reported as the analysis was based on a confidential net price) [
49]. In the present study, when attempting to mirror the CGRP analyses as closely as possible (patients with ≥3 previous treatments, 10-year time horizon, 30% stopping rule and using utility values mapped from MSQ to EQ-5D) the difference in QALYs for Botox vs placebo were 0.209 and with a corresponding ICER of EUR 16,625 in Sweden and 0.203 QALY difference with a corresponding ICER of 18,462 in Norway.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.