All of the 466 participants who enrolled completed the e-survey. The participants had an average age of 43 years (range 18–77), and 66% were women. A statistically significant difference emerged for age between the two genders, with males being older than females (45 years vs. 41 years). Working respondents reported that they had a full-time job in most cases (72%). The annual income ranges were equally distributed over the total number of respondents. Only 7% of respondents preferred to not declare their income. Statistically significant differences emerged between the genders for declared income, with lower incomes for females, who also reported less remunerative working roles.
Considering the characteristics of migraine, females showed a greater and significant duration of attack than males (duration greater than or equal to 4 hours: 70% vs. 49%). Aura affected more females than males (45% vs. 35%).
The socio-demographic characteristics of the sample, including disease characteristics, are summarized in Table 2.
Table 2
– Socio-demographic characteristics of the sample of respondents with the characteristics of migraine
Parameter | Total population | Males | Females | P-value |
Gender | 466 | 159 (34%) | 307 (66%) | |
Mean age (years) | 43 (18–77) | 45 | 41 | < 0.0001 |
Education level |
Primary school | 8% | 6% | 9% | 0.62 |
High school diploma | 52% | 54% | 51% |
Bachelor’s degree | 14% | 12% | 15% |
Master’s degree | 21% | 22% | 20% |
Doctorate | 6% | 7% | 5% |
Professional activity |
White collar | 62.00% | 78.60% | 53.50% | < 0.0001 |
Blue collar | 8.80% | 12.60% | 6.80% |
Retiree | 2.60% | 2.50% | 2.60% |
Homemaker | 13.50% | 1.90% | 19.50% |
Student | 5.40% | 0.60% | 7.80% |
Unemployed | 7.70% | 3.80% | 9.80% |
Income ranges declared by the workers (annual net) |
Less than €15.000 | 21% | 10% | 29% | < 0.0001 |
€15.000 - €19.999 | 21% | 14% | 26% |
€20.000 - €29.999 | 30% | 36% | 26% |
€30.000 or more | 28% | 40% | 19% |
Duration of migraine attack, frequency and symptoms |
Few minutes | 2% | 4% | 2% | < 0.0001 |
Up to 3 hours | 35% | 47% | 28% |
From 4 to 24 hours | 43% | 41% | 45% |
2–3 days | 20% | 8% | 25% |
Average number of attacks per month | 7.2 | 7.5 | 7.0 | 0.85 |
Number of attacks per month from 4 to 8 | 76% | 72% | 78% | 0.339 |
Number of attacks per month from 9 to 15 | 19% | 23% | 18% |
Number of attacks per month higher than 15 | 5% | 6% | 5% |
Presence of aura | 42% | 35% | 45% | 0.037 |
Regarding DCE, the participants reported 9,320 responses (20 questions each), with a significantly high number of cases (38%) in which respondents chose the opt-out option. In addition, 25% of respondents reported having encountered moderate to extreme difficulty in choosing among the alternatives.
The analysis of dominant preferences revealed that only 8 respondents expressed dominant preferences (score = 20) for the presence of adverse events, 7 for speed of effect and 7 for cost. The dominant preferences of other attributes were limited. The overall pattern of results suggests that the presence of side effects, speed of effect and the cost born by the patients were the most important factors for respondents in deciding which treatment they would select.
The full results of mixed logit models are presented in Table 3.
In Model 1, all the attributes significantly impacted the probability of choosing an alternative (p-values < 0.05). The negative sign of the coefficient for speed of effect (-0.00328) indicates that as the time to obtain a treatment effect increased, the patients’ likelihood of choosing this scenario decreased. The same results were found for adverse events and cost. In contrast, respondents preferred higher levels for strength of efficacy and duration of effect. Only the presence of adverse events showed a significant preference heterogeneity (standard deviation p-value = 0.003). These results are consistent with the indications received during the focus groups.
In Model 2, all the attributes (β1-β5) significantly impacted the probability of choosing an alternative. The presence of adverse events maintained heterogeneity in the preferences (standard deviation p-value = 0.007).
Table 3
Results of the mixed logit Models 1 and 2
Attributes | Model 1 | Model 2 |
Mean coefficient values | Standard deviations | Mean coefficient values | Standard deviations |
β | p-value | β | p-value | β | p-value | β | p-value |
Speed of effect (minutes) | -0.0033 | < 0.00001 | 0.0005 | 0.8500 | -0.0018 | 0.1830 | 0.0017 | 0.4440 |
Efficacy-strength (% of symptoms reduction) | 0.0242 | < 0.00001 | 0.0002 | 0.9030 | 0.0244 | < 0.00001 | 0.0002 | 0.9040 |
Duration of the effect (hours) | 0.0891 | < 0.00001 | -0.0008 | 0.9680 | 0.0903 | 0.0070 | -0.0003 | 0.9880 |
Presence of adverse events | -0.7580 | < 0.00001 | -0.2012 | 0.0030 | -0.4484 | < 0.00001 | -0.2404 | < 0.00001 |
Monthly cost born by the patient (€) | -0.0054 | < 0.00001 | 0.0009 | 0.5220 | -0.0046 | 0.0010 | 0.0020 | 0.1670 |
Female*speed of effect | | | | | -0.0037 | < 0.00001 | | |
Female*efficacy-strength | | | | | 0.0037 | 0.0670 | | |
Female*duration of the effect (hours) | | | | | 0.0237 | 0.1280 | | |
Female*presence of adverse events | | | | | -0.3346 | < 0.00001 | | |
Female*monthly cost born by the patient | | | | | 0.0012 | 0.0690 | | |
Age*speed of effect | | | | | 0.0000 | 0.5960 | | |
Age*efficacy-strength | | | | | 0.0000 | 0.7310 | | |
Age*duration of the effect (hours) | | | | | -0.0003 | 0.6850 | | |
Age*presence of adverse events | | | | | -0.0031 | 0.0660 | | |
Age*monthly cost born by the patient | | | | | 0.0000 | 0.1600 | | |
N. observations | 27,681 | 27,681 |
N. of respondents | 466 | 466 |
Log-likelihood | -8664 | -8550 |
Akaike information criterion (AIC) | 17,338 | 17,114 |
Gender interactions were statistically significant for speed of effect, presence of adverse events and cost. The results revealed that, compared to men, women had significantly higher preferences for quicker treatment effect and the presence of limited adverse events, and they reported higher preferences for more costly treatments. Age did not seem to influence the preferences of patients.
According to the Akaike information criterion, Model 2 showed a better fit than Model 1.
The WTP analysis performed on the base-case mixed logit model (Model 1) showed that respondents would be willing to pay €0.61 to anticipate the effect of the treatment by one minute, everything else being equal. Patients would be willing to pay €4.52, €16.66 and €141.68 to acquire a one percentage point increase in the strength of symptom reduction, for having an additional hour of effect duration and for de-escalating the adverse events, respectively, with everything else being equal. These results show that the reduction of adverse events is the most important dimension for which patients would be willing to pay the highest amount.
The WTP analysis performed separately on males and females showed a higher willingness to pay for women compared to men for all of the considered attributes (€0.98 vs. €0.18 to anticipate the effect of the treatment by one minute, €5.25 vs. €3.72 to gain a one percentage point increase in the strength of symptom reduction, €19.60 vs. €12.62 for having an additional hour of treatment effect duration, and €172.69 vs. €97.10 for de-escalating the adverse events).
The responses on the Likert scale were consistent with the results obtained in the DCE, and they confirmed the presence of adverse events as the most important treatment feature. The following other attributes, in order of importance, were efficacy-strength, speed of effect, duration of the effect and cost born by the patient (see Table 4).
Table 4
– Assessment of the importance of the considered attributes on the Likert scale
Grading | Speed of effect | Efficacy-strength | Duration of the effect | Presence of adverse events | Monthly cost |
Extremely important | 33.9% | 30.0% | 26.6% | 58.2% | 26.0% |
Very important | 33.0% | 40.6% | 37.8% | 22.3% | 30.5% |
Quite important | 28.5% | 27.5% | 31.1% | 14.6% | 32.6% |
Not very important | 4.1% | 1.7% | 3.9% | 4.3% | 8.6% |
Not at all important | 0.4% | 0.2% | 0.6% | 0.6% | 2.4% |