Patient Preferences Toward CGRP-Targeting Preventive Migraine Treatments in Japan: Discrete Choice Experiment

Abstract

Introduction

To date, no Japanese study on migraine has evaluated how treatment characteristics affect patient preferences between novel gepants and subcutaneous or novel intravenous calcitonin gene-related peptide (CGRP) monoclonal antibodies. A survey using a discrete choice experiment (DCE) was therefore conducted to identify treatment preferences for CGRP-targeting preventive therapies in this population.

Methods

Preventive-treatment attributes and levels were identified via a literature review, pivotal clinical trials, and international regulatory documents. These levels included those of CGRP-targeting therapies under development in Japan. A five-patient focus group was conducted to confirm patient understanding of the attribute descriptions and identify other potential attributes. Results from the focus group interview and from clinician input informed a questionnaire, which was revised following a pilot survey (N = 39), then an online main survey was conducted in May–June 2025 with eligible patients. A DCE was used to quantify patient preferences, and a conditional logit model was used to calculate preference weights and 95% confidence intervals.

Results

The questionnaire included six attributes: 50% reduction rate of migraine frequency, speed of effect, effect on impairment of daily activities, frequency of side effects, frequency of injection-site reactions, and administration method. The main survey included 324 patients (79% women; mean age 47 years) who experienced an average of eight migraine days per month. The survey revealed that effect on impairment of daily activities was the most important attribute, followed by administration method, and frequency of injection-site reactions. Regarding attribute levels, oral CGRP-targeting formulations were preferred over injectables within the specific attribute combinations used in this DCE.

Conclusions

These findings suggest that, under the design conditions used in this DCE, participants with migraine in Japan favored non-invasive oral CGRP-targeting treatments that effectively reduce impairment of daily activities.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s40120-026-01015-w.

Key Summary Points

Why carry out this study?

In Japan, the availability of calcitonin gene-related peptide (CGRP)-targeting preventive therapies for migraine is expanding beyond subcutaneous monoclonal antibodies (mAbs) to include oral gepants and intravenous formulations

Although consideration of patient preferences is desirable for shared decision-making, to date no study involving Japanese patients has evaluated how treatment attributes affect preferences when comparing novel gepants with both subcutaneous and novel intravenous CGRP mAbs

This study conducted a survey using a discrete choice experiment (DCE) to identify which treatment attributes matter most for CGRP-targeting preventive therapy among patients with migraine in Japan

What was learned from this study?

In this survey of 324 patients with migraine in Japan, the most important treatment attribute was improvement in impairment of daily activities (35.7%), followed by administration method (25.6%), and frequency of injection-site reactions (15.0%)

Patients preferred oral CGRP-targeting therapies over injectable options within the treatment profiles used in this DCE

These findings suggest that non-invasive, oral CGRP-targeting therapies that improve daily activities align better with patient preferences and support patient-centered migraine care

Introduction

Migraine is characterized by unilateral, pulsating, and severe headaches, and its prevalence in Japan is estimated to be approximately 8% [1, 2]. Migraine is more prevalent in women than in men, affecting approximately 20% of women in their 30s and 18% in their 40s [1]. According to the Global Burden of Disease Study, migraine ranks among the top three diseases in terms of age-standardized disability-adjusted life years (DALYs), predominantly due to reduced quality of life and disability [3]. Given the physical and psychological burden associated with migraine reported in Japan [4], effective management is crucial for alleviating the burden on patients.

According to the Clinical Practice Guideline for Headache Disorders 2021 in Japan (hereinafter, the Clinical Practice Guideline), acute treatment aims to rapidly resolve migraine attacks [5]. In addition, preventive treatment is recommended when impairment of daily activities due to migraine cannot be adequately managed by acute treatment alone, or when acute medication cannot be used [5]. Specifically, consideration of preventive treatment is recommended for patients who experience migraine attacks two or more times per month, or headaches causing impairment of daily activities on ≥ 3 days per month [5].

In Japan, a subcutaneous formulation of monoclonal antibodies (mAbs) targeting calcitonin gene-related peptide (CGRP) or the CGRP receptor—specifically galcanezumab—was approved and launched in April 2021 as a novel preventive treatment for migraine, followed by the mAbs fremanezumab and erenumab in August 2021. These agents are indicated for patients with migraine who experience impairment of daily activities despite having undergone non-pharmacological therapies or non-specific pharmacotherapies for migraine prevention, such as beta-blockers or antiepileptic drugs, without obtaining sufficient effects [6]. Additionally, the oral CGRP receptor antagonists rimegepant [7, 8] and atogepant [9–11] were launched in December 2025 and April 2026, respectively, shortly after the present study’s survey was conducted. Regulatory approval applications are also currently underway for an intravenous CGRP mAb [12, 13]. Although the mechanism of action of these CGRP-related agents (hereinafter, CGRP-targeting therapy) is considered to be fundamentally similar in terms of inhibiting CGRP function, various characteristics differ, including onset of action, administration method, and dosing frequency, in addition to efficacy and side effects [14, 15]. It is expected that if patient preferences are considered by physicians when selecting CGRP-targeting therapy from among these options, patient satisfaction with treatment and adherence will be enhanced, leading to better control of migraine symptoms [16–18].

In a previous Japanese study evaluating patient preferences using a discrete choice experiment (DCE), Seo et al. evaluated the preferences of patients with migraine and prescribing physicians regarding attributes of self-injectable preventive treatment, such as injection duration and needle removal [19]. In another study, Takeshima et al. reported on the relative importance of attributes, including method of delivery and reduction of pain, in a survey covering both acute and preventive treatments [20]. However, to the best of our knowledge, no study involving patients with migraine in Japan has yet evaluated how treatment attributes affect patient preferences when comparing novel gepants with both subcutaneous and novel intravenous CGRP mAbs.

Therefore, the objective of this study was to identify treatment preferences for CGRP-targeting therapy among patients with migraine in Japan using a DCE.

Methods

Study Design

This study was conducted in four steps (literature review; interviews; pilot survey; main survey) in accordance with the Good Research Practices for conjoint analysis and DCE established by the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) [21–23]. While DCE is widely used in economics to investigate individual preferences and trade-offs among product attributes, this method is also used in healthcare to quantify treatment preferences of patients and physicians across various disease areas [21, 22, 24].

Literature Review

A targeted literature review was performed using the MEDLINE and Ichushi-Web databases to comprehensively identify attributes and levels associated with migraine preventive treatment by searching for literature on treatment preferences (Search dates: 5 August 2024, and 7 October 2024). Information regarding candidate attributes and levels was extracted from identified relevant literature. Additionally, efficacy and safety data for CGRP-targeting therapies were obtained from pivotal clinical trials [7, 9, 10, 12, 13, 25–36] and usage precautions were derived from international regulatory approval documents [37–40]. Based on these sources, a draft list of attributes and levels was generated. Details on the initial candidate attributes and levels, together with the rationales for selection, are presented in Electronic Supplementary Material (ESM) Table S1.

Interviews

A focus group interview involving five patients with migraine was conducted on 29 March 2025. This qualitative format was chosen to facilitate discussion among participants and explore their views in depth. Details of the eligibility criteria for interviewees, screening sources, and screening methods are provided in ESM Method file 1. The objective was to elicit patient perspectives on the draft list of attributes and levels related to CGRP-targeting therapies. Specifically, the discussion focused on the perceived importance of attributes, the clarity of the wording used for attributes and levels, the necessity of adding any missing attributes or levels, and whether the attribute levels represented a realistic range for trade-offs. The interview was conducted online using a pre-developed semi-structured interview guide. The session was audio-recorded and transcribed verbatim, with all personally identifiable information anonymized. The comments of participants were categorized as either positive or negative to assess the importance of each attribute and level. Based on these findings, the wording of the attributes and levels was refined. Also based on the findings from the focus group interview, a screening process was conducted, and the number of attributes was adjusted to six for the draft questionnaire. Attributes were excluded if they: (1) were considered of low importance to patients; (2) were expected to have minimal impact on preferences for novel CGRP-targeting therapies; (3) were difficult to evaluate uniformly due to variations in perceived importance based on patient characteristics (e.g., disease severity); (4) exhibited conceptual overlap or similarity; or (5) raised concerns regarding uncertainty in level setting due to insufficient evidence available at the time to clearly define levels. This reduction was implemented because an excessive number of attributes in a DCE can increase cognitive burden on the participants, potentially compromising response accuracy and reliability [21, 41]. Furthermore, previous DCE studies evaluating patient preferences have typically used an average of approximately six attributes [42]. The attributes and levels included in the draft questionnaire were reviewed by two clinical experts with extensive experience in migraine management and the prescription of CGRP-targeting therapies (HI and HO) to ensure clinical relevance. The draft questionnaire was then fixed for the pilot survey.

Pilot Survey

An online pilot survey was conducted between 2 May and 7 May 2025, involving a small cohort of patients with migraine (N = 39) using the draft questionnaire containing 20 choice sets. The objectives were to collect feedback on the clarity of the instructions and wording, and to identify and address potential issues anticipated during the data collection process, such as, for example, the re-evaluation of eligibility criteria due to insufficient enrollment, adjustment of recruitment targets to mitigate demographic bias, and technical issues with display logic within the survey interface.

Main Survey

The questionnaire was refined and finalized by incorporating feedback from the pilot survey. Subsequently, the main survey was administered to the target number of patients with migraine between 28 May and 3 June 2025.

Eligibility Criteria

In accordance with the Clinical Practice Guidelines, approved labeling, and the local clinical context in Japan [5], eligibility criteria for the focus group interview and questionnaire surveys were defined as follows: (1) age ≥ 18 years at the time of the interview or survey; (2) self-reported physician diagnosis of migraine within the past year, with symptoms causing impairment of daily activities for ≥ 3 days in the preceding month; (3) use of acute and/or preventive treatment at the time of the interview or survey; and (4) age at migraine onset < 50 years. Individuals who participated in the focus group interview or the pilot survey were excluded from the main survey.

Patient Recruitment and Sample Size Estimation

Interview participants were recruited from a large-scale panel in Japan maintained by the marketing research agency ASMARQ Co., Ltd. (Tokyo, Japan), whereas the participants in the questionnaire surveys were recruited from a large-scale panel in Japan maintained by the healthcare marketing intelligence and evidence provider Intage Healthcare Inc. (Tokyo, Japan); both panels include individuals with self-reported migraine. Recruitment emails were distributed to these individuals, taking into account the distribution of age, sex, and residential region to ensure a balanced sample. The surveys were administered to those who accessed the recruitment site, met the eligibility criteria, and provided informed consent.

There is no standardized method for estimating the minimum sample size for DCE surveys; consequently, sample size determination is often based on previous studies [43]. Prior DCE research evaluating patient preferences has demonstrated sufficient clinical and statistical validity with 150 to 300 respondents [44, 45]. Furthermore, a meta-simulation study [43] recommends a sample size of 300 for designs with a structure similar to that of this study (i.e., 6–8 attributes, 3–4 levels per attribute, and 8–12 choice tasks per respondent). Therefore, we aimed to obtain at least 300 valid responses.

DCE Survey

At the beginning of the questionnaire, a hypothetical scenario was presented describing a patient who requires preventive treatment for migraine and is offered a new therapy by a physician. Participants were instructed to evaluate treatment profiles within this specific context (Table 1). Subsequently, participants were presented with a series of choice sets, each comprising two hypothetical treatment profiles formed by combining levels from each attribute, and were asked to select their preferred profile. A total of 17 choice sets were generated based on a D-efficient design with zero priors, using Ngene software version 1.4.0 (ChoiceMetrics, Sydney, Australia). All sets were presented to every participant without blocking. This design method optimizes the survey by minimizing the standard errors of the parameter estimates (D-error), thereby maximizing the statistical efficiency of the data collected [46]. Additionally, one choice set was repeated to assess the internal consistency of the responses from each participant. To enhance the understanding of the attribute levels by participants, the choice sets were presented with illustrations (Fig. 1). During the generation of choice sets, the following constraints were applied: (1) to reflect clinical practice, the oral medication level in the administration method attribute was paired exclusively with the not site reactions level in the frequency of injection-site reactions attribute; (2) choice sets containing a dominant profile were excluded; and (3) level overlap between the two profiles was avoided for each attribute. This constraint between oral medication and no injection-site reactions was introduced to avoid implausible combinations and improve the participants’ responses in the choice tasks [23]. As a result, the preference weights associated with oral administration should be interpreted within the context of this bundled profile structure rather than as an entirely independent estimate of the route of administration alone. The presentation order of the choice sets and the left–right positioning of the profiles were randomized for each participant to avoid potential bias due to learning effects and fatigue. Prior to the choice tasks, descriptions of the attributes and levels and DCE practice questions were presented to facilitate the understanding of participants. Additionally, sociodemographic and clinical characteristics were collected.

Table 1.

Hypothetical health conditions recalled by participants

Hypothetical health conditions recalled by participants
1. Migraines occur ≥ 4 days per month
2. Although lifestyle changes such as sleep and dietary habits and acute treatment for migraine are being implemented, migraines are interfering with daily activities such as family interactions and work
3. Preventive treatment for migraines was in use, but limited efficacy was observed, or continuation was not possible due to the occurrence of side effects

Fig. 1.

Example choice set. Participants were presented with a series of choice sets, each comprising two hypothetical treatment profiles defined by varying attribute levels, and were asked to select their preferred profile. Illustrations accompanied the choice sets to enhance the understanding of the attribute levels by participants. IV Intravenous, SC subcutaneous

Statistical Analyses

For sociodemographic and clinical characteristics, continuous variables were summarized as the mean and standard deviation (SD), while categorical variables were expressed as frequencies and percentages. The completion time for the choice sets was also collected and summarized using descriptive statistics.

Primary Analyses

Preference weights and their 95% confidence intervals (CIs) for each attribute level were estimated using a conditional logit regression model, based on effects-coded data from the responses. Conditional logit regression was used as the primary analytic approach to estimate average sample-level preferences. Although this standard DCE model provides interpretable preference weights and allows relative importance to be derived, it assumes homogeneous preferences and a constant scale across respondents [22, 23]. Additionally, the McFadden pseudo-R², log-likelihood, and Akaike information criterion (AIC) were calculated to assess model fit.

The relative importance of each attribute was calculated by dividing the range of preference weights (the difference between the maximum and minimum values) for that attribute by the sum of the ranges across all attributes, multiplied by 100, as shown in the following formula:

$$\text{Relative}\text{importance}\left(\text{\%}\right)=\frac{\text{Maximum}-\text{minimum}\text{preference}\text{weight}\text{for}\text{the}\text{attribute}}{\text{Sum}\text{of}(\text{maximum}-\text{minimum}\text{preference}\text{weight}\text{for}\text{all}\text{attributes})}\times 100.$$

The 95% CIs for the relative importance were estimated using the delta method.

Subgroup Analyses

Subgroup analyses were conducted to assess the consistency of preferences for preventive treatment across sociodemographic and clinical subgroups and to explore potential preference heterogeneity. Analyses were stratified by the following categories: sex; age (10-year increments); migraine type (episodic: < 15 disabling migraine days/month vs. chronic: ≥ 15 disabling migraine days/month); number of acute treatments in the past month (< 15 times vs. ≥ 15 times); current treatment (acute treatment, preventive treatment, or acute and preventive treatment); experienced with CGRP mAbs treatment (yes vs. no); and duration of illness (< 10 years vs. 10–19 years vs. 20–29 years vs. > 30 years). The analytical methods used were identical to those of the primary analyses.

Sensitivity Analyses

Sensitivity analyses were conducted to assess the robustness of the results using three subgroups: (1) migraine screener-positive participants; (2) participants with consistent responses; and (3) participants with reliable responses. For the first subgroup (migraine screener), a validated Japanese migraine screener was used [47]. Participants were classified as positive if they responded sometimes or half the time or more to at least two of the following four items: aggravation by routine physical activity or avoidance; nausea/stomach discomfort; photophobia; and disability. For the second subgroup (response consistency), the analysis was restricted to participants who selected the identical treatment profile in the repeated choice set. For the third subgroup (response reliability), participants meeting either of the following criteria were considered to have low reliability and were excluded: (1) completion of the entire questionnaire within 5 min; or (2) selection of the same alternative across all DCE tasks (i.e., straight-lining; consistently selecting either alternative A or alternative B).

For the analyses, only complete data from participants who provided informed consent were included. There were no missing values, as the web-based survey was designed to require a response for all items. Data from the pilot survey and the repeated choice set used to verify internal consistency were excluded from analyses.

All statistical analyses were performed using SAS Version 9.4 (SAS Institute, Cary, NC, USA).

Ethics

This study was conducted in accordance with the Ethical Guidelines for Medical and Biological Research Involving Human Subjects and the Declaration of Helsinki of 1964 and its later amendments. The study protocol was reviewed and approved by the Ethics Review Committee of Shiba Palace Clinic (Approval Number: 158252_rn-39651; Approval Date: 27 February 2025). Informed consent was obtained from all patients with migraine participating in the focus group interview, pilot survey, and main survey prior to the initiation of the investigation. All study participants were compensated for the time they spent on the study.

Results

Attributes and Levels Development

The targeted literature review identified 12 relevant articles. Of these, six studies evaluated preferences for migraine preventive treatment using a DCE [19, 20, 48–51], while six assessed preferences using interviews or questionnaires [52–57]. Based on these 12 articles, pivotal clinical trials of CGRP-targeting therapies, and international regulatory approval documents, a draft list of attributes and levels was developed (ESM Table S1). In total, 11 candidate attributes were identified: five related to efficacy, three related to safety, two related to product characteristics (administration method and storage conditions), and one related to out-of-pocket cost.

Five patients aged 23 to 52 years, consisting of four women and one man, were included in the focus group interview. All participants received oral acute treatments: two used only non-steroidal anti-inflammatory drugs (NSAIDs), one used NSAIDs with acetaminophen, one used NSAIDs with triptans, and one used only triptans. Two participants received oral preventive treatments: one used calcium antagonists, and the other used a Chinese medicine named tokishakuyakusan (ESM Table S2) [5, 58].

During the focus group interview, some participants suggested additional attributes, including palatability (e.g., taste) and duration of effect while on medication; however, these were not included in the final attribute set because they met the exclusion criteria. After the focus group interview, reduction in the number of migraine days was excluded from the candidate efficacy attributes; this exclusion was based on feedback suggesting that patient evaluations within the same attribute level would lack consistency, as the perceived value of a reduction depends on baseline disease severity. The potential heterogeneity of severity within the study population was also a contributing factor. Although some comments prioritized concomitant use with other medications and dose adjustment, these were not included. Considering patient background factors such as comorbidities and severity, these decisions were primarily determined by the prescribing physician rather than being solely by patient preference. Although concerns regarding out-of-pocket costs were raised, this attribute was also excluded. Because novel CGRP-targeting therapies (i.e., oral gepants and intravenous mAb) were unapproved in Japan at the time of the survey, it was not possible to present a basis for realistic costs, which would likely introduce uncertainty into participants' decision-making. Detailed information on the reasons for excluding each attribute from the candidate set is provided in ESM Table S3. Based on the focus group interview findings, along with clinical and methodological considerations, six attributes were selected for the pilot survey questionnaire: 50% reduction rate of migraine frequency; speed of effect; effect on impairment of daily activities; frequency of side effects; frequency of injection-site reactions; and administration method.

Based on participant comments from the pilot survey, the response burden of completing 20 choice sets was deemed high. To address this, the number of levels for frequency of injection-site reactions was decreased from four to three, and those for administration method reduced from five to four (ESM Table S4). The attributes and levels were then finalized for the main study questionnaire (Table 2). The full design matrix for the 17 choice sets used in the main survey is provided in ESM Table S5.

Table 2.

Attributes and levels

Attribute	Description	Level
50% reduction rate of migraine frequency	The proportion of patients whose monthly migraine frequency is reduced by 50%	• 20 patients in 100a • 40 patients in 100 • 60 patients in 100
Speed of effect	The time it takes to realize the effects of migraine prevention	• 1 montha • 1 week • 1 day
Effect on impairment of daily activities	How well does the treatment improve the interference with daily activities (family interactions, work, etc.) caused by migraine	• Not significantly improveda • Improved but still slightly impaired • Improved without any problems
Frequency of side effects	Risk of having side effects (nausea, dizziness, drowsiness, abdominal pain, constipation, etc.)	• 10 patients in 100a • 5 patients in 100 • 1 patient in 100
Frequency of injection-site reactions	Risk of having injection-site reactions (pain, redness, swelling, itching, etc.)	• 30 patients in 100a • 15 patients in 100 • Not site reactionsb
Administration method	Route and frequency of medication or injection administration	• Oral medication every dayab • Oral medication every other dayb • SC injection every 4 weeks • IV injection every 12 weeks

IV intravenous, SC subcutaneous

^aReference level

^bDue to constraints applied in the study to avoid implausible combinations of levels, the levels of oral medication every day and oral medication every other day were presented only in combination with the no site reactions level

Participant Characteristics

For the main survey, invitation emails were sent to approximately 30,000 individuals. Of the 648 candidates who accessed the recruitment site, 349 eligible patients provided informed consent. The questionnaire was completed by 324 patients (Fig. 2).

Fig. 2.

Flow chart of participants in the main survey. From a panel of 40,537 registered migraine patients, 30,395 were invited to participate in the survey, and 648 responded. Of the 349 respondents who met the eligibility criteria and consented to participate, 324 completed the main survey

The mean age of the 324 participants who completed the main survey was 47 years, and 255 (78.7%) were female. In terms of employment status, 34.6% were full-time employees, and 27.2% were part-time employees. Among these participants, 86.1% had episodic migraine, with a mean duration of illness of 23.1 years. In addition, 83.6% met the positive criteria for the migraine screener. Regarding acute treatment, 97.2% of participants used some form of medication, primarily NSAIDs, triptans, and/or acetaminophen. Preventive treatment was used by 50.3% of the participants, comprising mainly calcium channel blockers, antiepileptic drugs, and antidepressants. Additionally, approximately 10% received subcutaneous CGRP mAbs (Table 3).

Table 3.

Participant characteristics

Participant characteristics	Values (N = 324 participants)
Age (years), mean (SD)	47.2 (10.3)
Sex, n (%)
Male	66 (20.4)
Female	255 (78.7)
Other	0 (0.0)
No answer	3 (0.9)
Educational background, n (%)
Graduate school	14 (4.3)
College	100 (30.9)
Junior college	53 (16.4)
Vocational school	45 (13.9)
Senior high school	97 (29.9)
Junior high school	12 (3.7)
No answer	3 (0.9)
Employment status, n (%)
Full-time employee/company executive	112 (34.6)
Part-time employee/temporary worker	88 (27.2)
Self-employee	21 (6.5)
Other	2 (0.6)
Housework, students, seeking a job or retired	97 (29.9)
No answer	4 (1.2)
Marital status, n (%)
Unmarried	114 (35.2)
Married	156 (48.1)
Widowed	8 (2.5)
Divorced	42 (13.0)
No answer	4 (1.2)
Residential region, n (%)
Hokkaido, Tohoku	32 (9.9)
Kanto	122 (37.7)
Chubu	48 (14.8)
Kinki (Kansai)	54 (16.7)
Chugoku, Shikoku	24 (7.4)
Kyushu, Okinawa	44 (13.6)
Age at disease onset (years), mean (SD)	22.5 (10.0)
Duration of illness (years), mean (SD)	23.1 (13.1)
Frequency of migraine in the past month (times/month), mean (SD)	15.7 (16.1)
Number of experiencing disabling migraine in the past month (days/month), mean (SD)	7.9 (6.3)
Duration of attacks in the past month (hours/time), mean (SD)	7.7 (10.6)
Migraine type, n (%)
Episodic migraine (< 15 disabling migraine/month)	279 (86.1)
- Low-frequency episodic migraine (< 8 disabling migraine/month)	207 (63.9)
- High-frequency episodic (8–14 disabling migraine/month)	72 (22.2)
Chronic migraine (≥ 15 disabling migraine/month)	45 (13.9)
Test positivity rate according to the migraine screener, n (%)
Yes	271 (83.6)
No	53 (16.4)
The type of medical institution regularly visited for migraine treatment, n (%)
Clinic	235 (72.5)
General hospital	67 (20.7)
University hospital	22 (6.8)
Regularly consult with a headache specialist certified by the JHS, n (%)
Yes	102 (31.5)
No	66 (20.4)
Unknown	156 (48.1)
Use of acute medications in the past 3 months, n (%)a
Any acute treatment medication	315 (97.2)
- Oral medication/suppository	313 (96.6)
- Nasal spray	21 (6.5)
- Injection	27 (8.3)
No acute treatment in the past 3 months	9 (2.8)
Types of acute medications used in the past 3 months, n (%)a
Oral medication/suppository
- NSAIDs	171 (52.8)
- Triptan	168 (51.9)
- Acetaminophen	85 (26.2)
- Ditan	18 (5.6)
- Ergotamine	14 (4.3)
- Other analgesics	17 (5.2)
- Other
- - Chinese medicine	3 (0.9)
- - Medications other than those listed above	9 (2.8)
Nasal spray
- Triptan	21 (6.5)
Injection
- Triptan	19 (5.9)
- Other
- - Medications other than those listed above	8 (2.5)
Use of preventive medications in the past 3 months, n (%)a
Any preventive migraine medication	163 (50.3)
Oral medication	156 (48.1)
Injection	34 (10.5)
No preventive treatment in the past 3 months	16 (4.9)
Types of preventive medications in the past 3 months, n (%)a
Oral medication
- Calcium antagonists	71 (21.9)
- Antiepileptics	45 (13.9)
- Antidepressants	43 (13.3)
- Beta blockers	21 (6.5)
- Other
- - Dimetotiazine	9 (2.8)
- - Chinese medicine	16 (4.9)
- - Medications other than those listed above	22 (6.8)
Injection
- Anti-CGRP antibody, anti-CGRP receptor antibody	31 (9.6)
- Other	3 (0.9)
Number of oral preventive medications taken (tablets/day), mean (SD)	2.5 (3.4)
Frequency of preventive injection administrations (Once every X weeks), mean (SD)	3.6 (2.7)

CGRP Calcitonin gene-related peptide, JHS Japanese Headache Society, NSAID non-steroidal anti-inflammatory drug, SD standard deviation

^aDuplicate answers allowed

Preference Weights and Relative Importance

The mean time to complete the 17 choice sets was 468.3 (SD 1715.4) s.

The results of the preference survey indicated that, with the exception of speed of effect, preference weights were higher for levels associated with improvements in migraine frequency and daily activities, as well as lower frequencies of side effects (Fig. 3; Table 4). For effect on impairment of daily activities, participants most preferred improvement without any problems level (preference weight 0.557; 95% CI 0.502–0.612). Regarding administration method, participants significantly preferred oral formulations of gepants over injectable CGRP-targeting ones in the choice scenarios presented since the 95% CIs between the formulations do not overlap. In terms of specific administration levels, oral medication every other day (preference weight 0.411; 95% CI 0.300–0.523) had a numerically higher preference weight than oral medication every day (preference weight 0.324; 95% CI 0.203–0.446). Similarly, subcutaneous injection every 4 weeks (preference weight − 0.286; 95% CI − 0.376 to − 0.195) had a numerically higher weight than intravenous injection every 12 weeks (preference weight − 0.450; 95% CI − 0.525 to − 0.375). However, these differences were not statistically significant. Participants also favored treatments free from injection-site reactions (preference weight 0.280; 95% CI 0.206–0.335). For speed of effect, the preference weight was numerically higher for the 1-week level (preference weight 0.123; 95% CI 0.072–0.174) than for the 1-day level (preference weight 0.029; 95% CI − 0.030 to 0.088), but no statistical difference was observed, indicating that these levels were comparable.

Fig. 3.

Preference weights. Preference weights reflect the relative utility of each level within an attribute. A higher weight indicates a higher preference, whereas a lower weight indicates a lower preference. IV intravenous, SC subcutaneous

Table 4.

Preference weights

Attributes	Levels	Preference weight	95% CI (lower)	95% CI (upper)
50% reduction rate of migraine frequency	20 patients in 100a	− 0.146	− 0.205	− 0.087
	40 patients in 100	− 0.023	− 0.071	0.026
	60 patients in 100	0.169	0.111	0.227
Speed of effect	1 month*	− 0.152	− 0.207	− 0.096
	1 week	0.123	0.072	0.174
	1 day	0.029	− 0.030	0.088
Effect on impairment of daily activities	Not significantly improveda	− 0.644	− 0.695	− 0.592
	Improved but still slightly impaired	0.086	0.035	0.138
	Improved without any problems	0.557	0.502	0.612
Frequency of side effects	10 patients in 100a	− 0.121	− 0.179	− 0.063
	5 patients in 100	0.037	− 0.017	0.091
	1 patient in 100	0.084	0.035	0.134
Frequency of injection site reactions	30 patients in 100a	− 0.225	− 0.292	− 0.158
	15 patients in 100	− 0.056	− 0.121	0.010
	Not site reactions	0.280	0.206	0.355
Administration method	Oral medication every daya, b	0.324	0.203	0.446
	Oral medication every other daya, b	0.411	0.300	0.523
	SC injection every 4 weeks	− 0.286	− 0.376	− 0.195
	IV injection every 12 weeks	− 0.450	− 0.525	− 0.375
Log-likelihood		− 2985.599
McFadden pseudo-R2		0.218
AIC		5997.197

AIC Akaike information criterion, CI confidence interval, IV intravenous, SC subcutaneous

^aReference level

^bDue to constraints applied in the study to avoid implausible combinations of levels, the levels of oral medication every day and oral medication every other day were presented only in combination with the no site reactions level

In terms of relative importance, effect on impairment of daily activities (35.7%; 95% CI 32.9–38.6%) was the highest, followed by administration method (25.6%; 95% CI 21.0–30.2%) and frequency of injection-site reactions (15.0%; 95% CI 11.4–18.6%). Conversely, frequency of side effects (6.1%; 95% CI 3.5%–8.8%) was the least important among the six attributes (Fig. 4).

Fig. 4.

Relative importance. Relative importance was calculated as the difference between the maximum and minimum preference weights for each attribute. The 95% confidence intervals (CIs) for relative importance were estimated using the delta method. Higher relative importance indicates that the attribute has a greater impact on the choices of participants

The results of all subgroup analyses indicated trends that were generally consistent with those of the overall study population. The relative importance of the effect on impairment of daily activities ranged from 25.8% to 40.9%, that for administration method range from 12.4% to 40.6%, and that for frequency of injection-site reactions ranged from 8.8% to 18.1% (ESM Figs. S1, S2). However, some subgroups showed slightly different trends in attribute levels compared with those for the overall population. In patients with prior experience using CGRP mAbs or with chronic migraine, the difference in preference weights between the formulations was lower than that in patients without such experience or with episodic migraine. Moreover, among subgroups of patients who had a high frequency of acute treatment (≥ 15 times/month), the preference level for daily administration was numerically higher than that for those with every-other-day administration.

Sensitivity analyses showed results consistent with those of the overall population regarding the range for the effect on impairment of daily activities (35.7–37.4%), the administration method (25.0–26.6%), and the frequency of injection-site reactions (12.4–15.5%) (ESM Figs S3, S4). Of the 324 participants included in the primary analyses, 53 participants (16.4%) did not test positive on the migraine screener, 67 participants (20.7%) failed the internal consistency check using the repeated choice set, and four participants (1.2%) were excluded based on the data-reliability criteria, including two for straight-lining and two for completing the questionnaire within 5 min.

Discussion

To our knowledge, this is the first study to evaluate how differences in treatment attributes affect patient preferences between novel gepants and both subcutaneous and novel intravenous CGRP mAbs among a panel-based migraine sample in Japan. This survey of 324 patients regarding preventive treatment preferences revealed that effects on impairment of daily activities and administration method were the most important attributes. Moreover, the study demonstrated that participants with migraine tended to favor oral formulations of gepants over injectable CGRP-targeting formulations within the specific attribute combinations used in this DCE. These findings may help to inform physicians regarding preventive treatment strategies based on the preferences of patients for CGRP-targeting therapies; in particular, when a new gepant or intravenous CGRP mAb becomes available in Japan. In addition, these findings may help to facilitate dialogue between physicians, other healthcare providers, and patients on preventive treatment options, which may in turn improve patient adherence and better control of migraine symptoms.

The sociodemographic characteristics of the participants in this survey, including sex, age, residential region, and employment status, were generally consistent with those reported in large-scale cross-sectional epidemiological studies involving patients with migraine [59, 60]. However, the proportion of participants receiving preventive treatment in this study was approximately 50%, which is higher than that reported in the previous studies. Although Takeshima et al. noted in a similar DCE survey that treatment strategies in Japan have recently shifted toward preventive treatment [20], the authors of a survey using a Japanese claims database (2018–2023) reported that approximately 20% of patients had experience with preventive treatment [61]. Similarly, the CaMEO-I study—an internet-based, cross-sectional observational study of patients with migraine in six countries (US, Canada, UK, Germany, France, and Japan)—reported that the proportion of patients in Japan receiving preventive treatment was 10.1%, which was the lowest of the six countries included in the study (average: 15.5%) [62]. Therefore, the higher rate observed in the present study suggests the possibility of selection bias related to the patient panel or survey methodology.

In the present study, preference was highest for the attribute effect on impairment of daily activities. Previous studies have similarly highlighted a strong preference for efficacy in preventive treatment among patients with migraine [48, 50, 63]. Whichello et al. conducted a preference survey involving US patients with episodic migraine, assessing attributes such as treatment administration, chance of a ≥ 50% reduction in monthly migraine days, onset of treatment effect, impact on daily activities, and reduction in acute medication use. These authors found that the preference for the chance of ≥ 50% reduction in monthly migraine days was higher than that for impact on daily activities [63]. Although their study also examined the differences in the attributes of both injectable and oral CGRP-targeting therapies, the attributes and level settings differed from those used in our study. Therefore, direct comparison of the attributes across studies should be interpreted with caution. Under the specific conditions presented in the present study, improvement in daily activities appeared to be an important consideration for participants.

Furthermore, regarding administration method, which was identified as the second most important attribute in this study, patients demonstrated a higher preference for oral formulations of gepants over injectable CGRP-targeting formulations in the choice scenarios presented. Although the CGRP mAbs approved in Japan at the time of the survey were injectable, these results suggest a potential demand for oral CGRP-targeting options among patients with migraine. Several previous studies have also reported a strong preference for attributes related to the mode of administration in CGRP-targeting therapies [20, 49, 64]. Lipton et al. reported that, among patients who are CGRP mAb-naive with episodic migraine in the USA, the administration method was the second most important attribute, following the avoidance of moderate-to-severe nausea [64]. Their study, which also evaluated the treatment preferences of patients for CGRP-targeting therapies—including gepants and CGRP mAbs—found that oral formulations of gepants were significantly preferred over injectable CGRP-targeting ones. Consistent with our findings, these results underscore that the administration method, particularly oral formulations of gepants, is a key driver of treatment selection. In contrast, other studies found that patients placed less importance on administration methods [50, 63]; notably, Whichello et al. reported that administration method had the lowest relative importance. In their study, the inclusion of a high proportion of patients with severe migraine likely led to a higher preference for efficacy-related attributes, resulting in a relatively lower importance for the administration method. Additionally, nearly half of their participants were current or previous users of self-injectable formulations. This suggests they may have had less resistance to injectables, making the preference gap between dosage forms less pronounced than in our study. Nevertheless, across all studies comparing injectable and oral CGRP-targeting formulations, the trend favoring oral formulations remained consistent. This suggests that patients with migraine potentially desire oral CGRP-targeting preventive treatments over injectables.

Although the results of the subgroup analyses were generally consistent with those of the overall population, some variations were observed in specific subgroups. Among patients with experience using subcutaneous CGRP mAbs, the preference for oral CGRP-targeting formulations over injectable ones remained consistent with the preference of the overall population; however, the difference in preference weights between dosage forms was smaller than that in the subgroup without such experience. Consequently, the relative importance of the administration method attribute was lower than that in the overall population. This may reflect reduced resistance to injectable formulations resulting from prior experience. Similarly, in subgroups with a high frequency of acute treatment (≥ 15 times/month) or long disease duration (≥ 30 years), the preference for oral CGRP-targeting formulations over injectable ones persisted, although the pattern of preference across oral dosing frequencies differed somewhat. Because some subgroup sizes were small and potential confounding factors cannot be excluded, these findings should be interpreted with caution. Further investigation is warranted.

Limitations

This study has several limitations. First, the survey was conducted among patients registered in a large-scale panel in Japan. Because patients with migraine who were not registered in the panel or who did not wish to register were not included, selection bias may have occurred. In addition, the response rate was low, and non-response bias may have influenced the results. Therefore, the sample should be interpreted as a treatment-engaged migraine sample rather than as representative of the national migraine population in Japan. Second, diagnostic information was based on self-reports of a diagnosis by a physician, introducing the potential for self-reporting bias. To address this, a sensitivity analysis restricted to patients who tested positive on a migraine screener was conducted, yielding results consistent with the primary analyses. Furthermore, because confirmation of current medication use was part of the eligibility criteria, the reliance on self-reported diagnoses likely did not significantly affect result interpretation. Third, to facilitate participant decision-making and ensure survey feasibility, the number of attributes characterizing preventive treatment was limited, and levels were simplified to three or four. However, actual clinical practice is more complex, and this simplification may have influenced the estimated preference weights and relative importance. In addition, because this was a hypothetical DCE, participants may not have fully considered all treatment trade-offs that would be discussed in routine clinical practice, such as adherence, safety-related issues, or pregnancy-related considerations. Moreover, oral medication was structurally linked with no injection-site reactions in the DCE design, so the estimated preference weights for administration method should be interpreted cautiously as reflecting the bundled profile rather than the route of administration alone. As a result, the relative importance assigned to administration method may partly reflect the bundled absence of injection-site reactions. Fourth, attributes and levels were established based on previous studies, clinical trial data for CGRP-targeting therapies available at the time, and patient insights from a focus group interview. Nevertheless, information bias may have occurred due to inconsistencies in sources regarding long-term efficacy and safety, or the inability to incorporate data unavailable at the time of the survey. In addition, relative importance in a DCE depends on the range of levels selected for each attribute. Therefore, the low relative importance of the frequency of side effects in the present study may partly reflect the relatively narrow range of side-effect frequencies presented to participants. Fifth, the primary analysis used a conditional logit model, which assumes the independence of irrelevant alternatives and homogeneous preferences across respondents. These assumptions may not have been fully met in our sample. Subgroup analyses suggested some variation in preference patterns across participant groups, indicating that underlying preference heterogeneity may not have been fully captured by the primary model. More flexible approaches, such as mixed logit or latent-class models, may be considered in future studies to better account for such variation [22, 23]. Finally, the DCE did not include all potentially relevant treatment characteristics, such as out-of-pocket cost. Although cost is often an important determinant in DCEs, it was not included because it may have introduced uncertainty into respondents’ decision-making; however, under Japan’s fixed-price reimbursement system, subcutaneous CGRP mAbs and newly launched gepants were ultimately associated with similar out-of-pocket costs (approximately JPY 12,000–13,000 per month), which may have limited the impact of omitting cost on respondents’ preferences [65, 66].

Further research incorporating more detailed attributes and levels may be necessary to better understand how the preferences of patients with migraine in Japan would be affected upon the actual introduction of novel CGRP-targeting therapies.

Conclusions

The results of this study indicate that participants with migraine in this Japanese survey tended to favor oral CGRP-targeting therapies over intravenous ones, particularly those that effectively reduced impairment of daily activities within the specific attribute combinations used in this DCE. The results may help physicians in promoting patient-centered care and shared decision-making, by ensuring that the importance of treatment attributes from the perspective of the patient are considered when selecting migraine preventive treatments.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Conceptualization: Toru Yamazaki and Keigo Hanada. Data curation: Keigo Hanada. Formal analyses: Keigo Hanada. Funding acquisition: AbbVie GK. Methodology: Toru Yamazaki, Keigo Hanada, Hiroyuki Hozawa, and Tatsunori Murata. Project administration: Toru Yamazaki. Software: Keigo Hanada. Supervision: Hisaka Igarashi and Toru Yamazaki. Validation: Hisaka Igarashi and Hitomi Onomura. Visualization: Toru Yamazaki and Keigo Hanada. Writing–original draft: Toru Yamazaki and Keigo Hanada. Writing–review & editing: Hisaka Igarashi, Toru Yamazaki, Keigo Hanada, Hitomi Onomura, Hiroyuki Hozawa, and Tatsunori Murata.

Funding

AbbVie GK funded this study and the journal’s Rapid Service Fee and Open Access Fees; it also participated in the study design, research, analyses, data collection, interpretation of data, review, and approval of the manuscript for publication. All authors had access to relevant data and participated in the drafting, review, and approval of this publication. No honoraria or payments were made for authorship.

Data Availability

AbbVie is committed to responsible data sharing regarding the studies we sponsor. This includes access to anonymized, individual, and study-level data (analysis data sets), as well as other information (e.g., protocols, clinical study reports, synopses, or statistical analysis plans), as long as the studies are not part of an ongoing or planned regulatory submission. These study data can be requested by any qualified researchers who engage in rigorous, independent, scientific research, and will be provided following review and approval of a research proposal, Statistical Analysis Plan (SAP), and execution of a Data Use Agreement (DUA). Data requests can be submitted at any time after approval in the USA and Europe and after acceptance of this manuscript for publication. The data will be accessible for 12 months, with possible extensions considered. For more information on the process or to submit a request, visit the following link: https://vivli.org/ourmember/abbvie/ then select Home.

Declarations

Conflict of Interest

Hisaka Igarashi has received consulting fees from Sawai Pharmaceutical Co., Ltd., AbbVie GK, Pfizer Inc.; and lecture honoraria from Otsuka Pharmaceutical Co., Ltd., Eli Lilly Japan K.K, Amgen K.K., Daiichi Sankyo Co., Ltd., and AbbVie GK. Toru Yamazaki, Hitomi Onomura and Hiroyuki Hozawa are employees of AbbVie GK and may hold AbbVie stocks/stock options. Keigo Hanada and Tatsunori Murata are employees of CRECON Medical Assessment Inc., which received funding from AbbVie GK to conduct the study. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Ethical Approval

This study was conducted in accordance with the Ethical Guidelines for Medical and Biological Research Involving Human Subjects and the Declaration of Helsinki of 1964 and its later amendments. The study protocol was reviewed and approved by the Ethics Review Committee of Shiba Palace Clinic (Approval Number: 158252_rn-39651; Approval Date: February 27, 2025). Informed consent was obtained from all patients with migraine participating in the focus group interview, pilot survey, and main survey prior to the initiation of the investigation.